Abstract

The superior colliculus (SC) is thought to be unresponsive to stimuli that activate only short wavelength-sensitive cones (S-cones) in the retina. The apparent lack of S-cone input to the SC was recognized by Sumner et al. [Sumner, P., Adamjee, T., & Mollon, J. D. Signals invisible to the collicular and magnocellular pathways can capture visual attention. Current Biology, 12, 1312–1316, 2002] as an opportunity to test SC function. The idea is that visual behavior dependent on the SC should be impaired when S-cone stimuli are used because they are invisible to the SC. The SC plays a critical role in blindsight. If the SC is insensitive to S-cone stimuli blindsight behavior should be impaired when S-cone stimuli are used. Many clinical and behavioral studies have been based on the assumption that S-cone-specific stimuli do not activate neurons in the SC. Our goal was to test whether single neurons in macaque SC respond to stimuli that activate only S-cones. Stimuli were calibrated psychophysically in each animal and at each individual spatial location used in experimental testing [Hall, N. J., & Colby, C. L. Psychophysical definition of S-cone stimuli in the macaque. Journal of Vision, 13, 2013]. We recorded from 178 visually responsive neurons in two awake, behaving rhesus monkeys. Contrary to the prevailing view, we found that nearly all visual SC neurons can be activated by S-cone-specific visual stimuli. Most of these neurons were sensitive to the degree of S-cone contrast. Of 178 visual SC neurons, 155 (87%) had stronger responses to a high than to a low S-cone contrast. Many of these neurons' responses (56/178 or 31%) significantly distinguished between the high and low S-cone contrast stimuli. The latency and amplitude of responses depended on S-cone contrast. These findings indicate that stimuli that activate only S-cones cannot be used to diagnose collicular mediation.

INTRODUCTION

Blindsight is an enigma. Patients with lesions in primary visual cortex (V1) are unable to describe or perceive stimuli in their blind field, yet they retain the ability to detect, localize, and discriminate unseen stimuli (Cowey, 2010; Weiskrantz, 2004; Sanders, Warrington, Marshall, & Wieskrantz, 1974). This perplexing phenomenon has been demonstrated in both humans and monkeys (Gross, Moore, & Rodman, 2004; Stoerig & Cowey, 1997; Cowey & Stoerig, 1995; Moore, Rodman, Repp, & Gross, 1995). Many researchers have investigated the capabilities and mechanisms of blindsight, yet the pathways involved remain only partially understood (Leopold, 2012).

Multiple cortical and subcortical structures contribute to blindsight. Projections from the superior colliculus (SC) to extrastriate cortex play a critical role in producing neuronal visual responses in the absence of V1 (Gross, 1991). The SC and extrastriate visual cortex are both crucial for blindsight (Cowey, 2010; Yoshida, Takaura, Kato, Ikeda, & Isa, 2008; Weiskrantz, 2004; Goebel, Muckli, Zanella, Singer, & Stoerig, 2001; Vaughan & Gross, 1969). A possible blindsight pathway requires the LGN of the thalamus. When the LGN is inactivated in awake, behaving monkeys that have V1 lesions, extrastriate activation and blindsight behavior are both abolished (Schmid et al., 2010). The SC and retina project to the LGN, which projects directly to extrastriate cortex. After removal of V1, this pathway remains intact and is vital to blindsight behavior (Leopold, 2012).

An important strategy in studies of blindsight in humans has been to block visual input to the SC psychophysically. It is widely believed that the SC does not receive input from short wavelength sensitive cones (S-cones) in the retina (Sumner, Adamjee, & Mollon, 2002). The reasoning is that, by using S-cone stimuli, visual pathways that depend on the SC should be blocked or impaired. Several studies have used S-cone stimuli to test whether the SC mediates blindsight in humans (Alexander & Cowey, 2010; Leh, Ptito, Schönwiesner, Chakravarty, & Mullen, 2010; Tamietto et al., 2010; Marzi, Mancini, Metitieri, & Savazzi, 2009; Leh, Mullen, & Ptito, 2006). Experimental paradigms have ranged from target detection to fMRI activation. The results have demonstrated selective impairments when S-cone stimuli are used as compared with luminance stimuli. The conclusion is that the SC is likely to be responsible for visual abilities demonstrated by blindsight patients.

Beyond blindsight, many behavioral studies have been based on the assumption that S-cone isolating stimuli do not drive cells in the SC. If this were true, the use of S-cone isolating stimuli would be a convenient way to assess collicular function. This strategy was first proposed in an influential behavioral study of SC function (Sumner et al., 2002). Since its inception, this technique has been used in a number of human psychophysical studies of the SC. The appeal of this method is that the SC can be “lesioned” without any damage to the brain, making it useful in healthy human volunteers. Many studies using S-cone stimuli have found differences in behavioral responses to S-cone as compared with luminance stimuli. When behavioral effects are observed, the interpretation is that the SC must play a critical role in the affected behavior.

The original idea that the SC is blind to S-cone isolating stimuli is based on anatomical and physiological studies. Early physiology studies found that SC neurons and the retinal ganglion cells that form the retinotectal pathway are not color opponent and are not sensitive to S-cones (De Monasterio, 1978b; Marrocco & Li, 1977; Schiller & Malpeli, 1977). The corticotectal pathway in turn carries information originating only from the magnocellular layers of the LGN. Neurons in the magnocellular layers carry summed inputs from long and middle wavelength sensitive cones (L- and M-cones), but not S-cones. Single neurons in monkey SC are silenced when the magnocellular layers of the LGN are inactivated (Schiller, Malpeli, & Schein, 1979). The conclusion is that SC neurons should be responsive to achromatic luminance stimuli but not to chromatic stimuli.

No single neuron studies have directly tested whether cells in the SC respond to individually calibrated S-cone isolating stimuli. Studies of SC afferents only make clear that these afferents are not color opponent and not whether SC neuronal responses are affected by calibrated S-cone isolating stimuli. One recent study reported neuronal sensitivity in the SC of macaques to colored stimuli (White, Boehnke, Marino, Itti, & Munoz, 2009). Another has reported the absence of S-cone input in the marmoset SC (Tailby, Cheong, Pietersen, Solomon, & Martin, 2012). Our goal was to test directly whether S-cone isolating stimuli can activate single cells in the SC of awake, behaving macaque monkeys under conditions that explicitly replicate those used in human behavioral and clinical research. Surprisingly, our data show that S-cone-specific stimuli produce robust activation of SC neurons.

METHODS

Two adult male rhesus monkeys (Macaca mulatta) were used in these experiments. Animals were cared for in accordance with National Institutes of Health guidelines. The University of Pittsburgh Institutional Animal Care and Use Committee approved all experimental protocols. Monkeys weighed 13 and 8.5 kg (monkey CA and monkey FS, respectively). Both had normal trichromatic color vision as evidenced by their ability to make saccades to L-cone and S-cone isolating stimuli (data not shown).

Each monkey underwent sterile surgery to implant an acrylic cap with an embedded head restraint bar, scleral search coils, and a recording chamber as described elsewhere (Dunn, Hall, & Colby, 2010). We used MRI to guide correct placement of the SC recording chambers, which were positioned posteriorly on the midline at an angle of 40°. The SC was located approximately 25–30 mm below the surface of the brain.

Data Acquisition

During experiments, animals sat in a primate chair with head fixed. Monkeys viewed a CRT monitor at a distance of 30 cm. Eye position was monitored using scleral search coils (Judge, Richmond, & Chu, 1980) with a Riverbend field driver and signal processing filter (Riverbend Instruments, Inc., Birmingham, AL). Eye position voltages were continuously monitored on-line through NIMH Cortex software (provided by Dr. Robert Desimone) and saved for off-line analysis of saccades on a separate computer running Plexon software at a sampling rate of 1000 Hz (Plexon, Inc., Dalls, TX). Neuronal spiking activity was recorded using tungsten microelectrodes (Frederick Haer, Bowdoinham, ME) inserted into the SC through stainless steel guide tubes stabilized in a nylon grid system (Crist Instrument Co., Hagerstown, MD). Voltage signals were amplified, filtered, and sorted using on and off-line template-matching and PCA (Plexon, Inc., Dalls, TX). The collicular surface was physiologically identified by the sudden appearance of light sensitive neuronal activity as the electrode was advanced. At deeper locations, eye movement-related activity was observed. Visual and motor receptive fields were in agreement with known SC maps (Goldberg & Wurtz, 1972; Schiller & Stryker, 1972). Trial types, events, and outcomes were monitored with Cortex software and stored with neuronal spike data at 40 kHz using Plexon hardware and software.

Stimulus Presentation and Control

Stimulus timing and presentation were controlled using NIMH Cortex software. Stimuli were presented on a 19-in. ViewSonic color CRT monitor at a refresh rate of 85 Hz using an 8-bit DAC with an ATI Radeon X600 SE graphics card. To generate stimuli, the monitor was calibrated for color and luminance using a Photo Research PR-655 SpectraScan spectroradiometer integrated with custom MATLAB (Mathworks, Natick, MA) software. The accuracy of the calibration was verified by independently measuring the spectral composition of all stimuli used. Stimulus colors were chosen from MacLeod–Boynton (MB) color space (MacLeod & Boynton, 1979) as defined using the Smith and Pokorny cone fundamentals (Smith & Pokorny, 1975).

Stimuli and Background

We presented stimuli defined by either luminance or S-cone contrast. In MB space, S-cone excitation is isolated along a vertical vector, perpendicular to the L- and M-cone excitation axis, known as the tritan vector. The actual tritan vector varies across individuals and spatial locations because of variations in cone density and macular pigment (Hall & Colby, 2013; Sumner et al., 2002; Snodderly, Auran, & Delori, 1984; Snodderly, Brown, Delori, & Auran, 1984). We presented S-cone stimuli chosen from psychophysically calibrated tritan vectors determined separately for each spatial location and each animal (Hall & Colby, 2013). Briefly, our calibration approach takes advantage of “transient tritanopia” (Smithson, Sumner, & Mollon, 2003; Mollon & Polden, 1975). The principle is that sensitivity in the S-cone opponent channel is briefly and selectively decreased after adaptation to a bright yellow background. Elevated L- and M-cone channel activity persists after excessive stimulation by the bright yellow background, which causes selective desensitization of S-cone opponent pathways. Elevated L- and M-cone activation means that greater S-cone excitation is needed to outweigh them and thus detect a true S-cone isolating stimulus. That is, detection threshold is increased for S-cone isolating stimuli. The increase in threshold is greatest for the stimulus that best matches the exact S-cone isolation point for a given viewer. The calibrated individual and location-specific tritan vector is defined as the vector in MB color space most desensitized after adaption to a bright yellow background.

The calibrated tritan vector for each monkey and spatial location examined was rotated to the right of the theoretical vertical tritan line in MB space. Calibrated tritan vectors were brought into congruence with standard MB space by linearly transforming their MB values such that the calibrated tritan vector extends vertically and orthogonal to the standard L- and M-cone excitation axis.

We were able to identify four location-specific tritan vectors in monkey FS and four in monkey CA using our procedure outlined above (Hall & Colby, 2013). Neurons in the SC with receptive fields corresponding to these calibrated locations were targeted in each animal, and all stimulus presentations took place at these calibrated locations.

Stimuli were presented on an equal energy gray (EEG) background of luminance noise (Sumner et al., 2002; Birch, Barbur, & Harlow, 1992). The background was a full screen array of 1° × 1° squares whose color was EEG but whose individual luminances changed at random every four monitor frames (∼47 msec). This flickering background removes potential artifacts created by stimuli that are not exactly equiluminant with the background. Stimuli presented on such a background are therefore detectable only through contrasts falling outside the range of the noise. The luminance of the squares was uniformly distributed across nine possible values ranging from 18.78 to 22.55 cd/m2, spaced in increments of approximately 0.5 cd/m2. The mean background luminance across all possible values was 20.73 cd/m2, with mean color coordinates in MB space of (0.66692, 0.01493) as compared with the theoretical value (0.66667, 0.01500).

The goal of our experiment was to test SC neuronal responses to S-cone-specific stimuli. We converted stimuli to DKL contrast space to make the effects of contrast in color opponent channels explicit (Brainard, 1996; Derrington, Krauskopf, & Lennie, 1984). The mean color and luminance of the background noise in MB space was used as the basis point to convert stimuli to DKL space. This procedure gives stimulus contrasts with respect to the mean color and luminance of the EEG noise background. Conversion to DKL space gives contrasts corresponding to the luminance mechanism (L + M), the L- and M-cone opponent mechanism (L − M) and the S-cone opponent mechanism [S − (L + M)], that is, the S-cone isolating direction. Following the procedure described by Brainard (1996), we normalized DKL space such that a stimulus isolating a specific mechanism in DKL space with unit pooled cone contrast (i.e., the length of the 3-D vector in cone contrast space equals one) would correspond to a contrast of 100%. This normalization procedure has the advantage that it is independent of observer and experimental setup. We define the DKL coordinate contrasts for each mechanism as a vector of the form (L + M, L − M, S − (L + M)). In these terms, the minimum contrast of the luminance noise background was (−16.2298, 0.0091, 1.0320), and the maximum was (15.3497, −0.0402, −0.2187).

Three stimulus types were presented on the background array of squares: luminance, high-contrast S-cone, and low-contrast S-cone. Stimuli were 1° × 1° squares embedded in the luminance noise background. The luminance contrast stimulus (24.92 cd/m2) isolates the luminance mechanism (35.1922, 0.1817, 0.5641). This stimulus is more than double the maximum luminance contrast of the noise background. The other two stimuli activated the S-cone opponent mechanism. One was a high S-cone contrast stimulus whose DKL coordinates across all three mechanisms ranged from (−5.4318, 0.0157, 95.7475) to (−8.7634, −0.1684, 95.9704). The other was a low S-cone contrast stimulus whose DKL coordinates ranged from (−2.7450, 0.0462, 28.3539) to (−6.2916, −0.2824, 29.5714). Both S-cone stimuli slightly decreased DKL luminance contrast and contained only small, inconsistent contrasts in the DKL L-M opponent mechanism. Such small and inconsistent deviations in mechanisms other than the S-cone opponent channels would not be expected to produce consistent effects on neurons. Luminance and L-M opponent contrasts remained within the range covered by the background noise. S-cone stimuli presented on the noisy background were therefore detectable only on the basis of S-cone contrast and are thus S-cone isolating stimuli. The high-contrast S-cone and luminance stimuli were chosen to be similar to those found in previous human psychophysics research. The low-contrast S-cone stimulus was chosen to lie near the luminance stimulus and far from the high-contrast S-cone stimulus in DKL space.

In 10 neurons, we additionally tested neuronal responses to an L-cone stimulus as defined in DKL space (data not shown). SC neurons were responsive to the L-cone stimulus, as found previously (White et al., 2009). This confirmed that our results are not specific to S-cone stimuli or a subpopulation of SC neurons, but that SC neurons can be broadly activated by different cone inputs.

Behavioral Tasks

We used a standard memory-guided saccade (MGS) task to identify neurons with receptive fields at locations at which tritan vectors had been calibrated (Hall & Colby, 2013). We tested each SC neuron in two tasks. We used the MGS task to classify neurons as visual or visuomotor. We then used a fixation task to measure neuronal responses to the luminance and to high- and low-contrast S-cone stimuli. During previous S-cone calibration sessions, S-cone stimuli were behaviorally relevant because they served as saccade targets for the animals (Hall & Colby, 2013). In the fixation task of these experiments, no response was required after stimulus presentation.

MGS Task

During MGS trials, the stimulus and fixation point were maximum monitor intensity white (63.15 cd/m2) presented on a dark background (<0.01 cd/m2). These stimuli represent the maximum contrast possible with our monitor. At the beginning of each trial, a 1° × 1° fixation cross appeared and the monkey fixated for 200–400 msec. While maintaining fixation, a 0.5° diameter circle was presented in the recorded neuron's receptive field for four frames (∼47 msec). Monkeys were required to maintain fixation for an additional 300–500 msec after stimulus presentation before the fixation cross was turned off, cueing the animals to make a saccade to the remembered location of the target. Saccades within approximately 1.5° of the target location were rewarded with a drop of liquid. Data for the MGS task were collected in a block of 20 trials before the fixation task and a block of 20 trials after the fixation task.

Fixation Task

At the beginning of each trial, a 1° × 1° black (<0.01 cd/m2) fixation cross was presented as the luminance noise-masking background was introduced. Animals fixated the central cross for 200–400 msec (Figure 1, left), after which one of the three stimulus types was presented. The stimuli appeared for four frames (∼47 msec) synchronously with the change in the luminance noise background (Figure 1, center). The figure shows the high-contrast S-cone stimulus, outlined in white for clarity. This outline was not present during the actual task. All three stimuli were relatively difficult for a human observer to detect. After stimulus presentation, animals were required to continue fixating for an additional 200–400 msec to receive a liquid reward. After the additional fixation period, the luminance noise and fixation cross were turned off and the intertrial interval (ITI) began (Figure 1, right). During the 400-msec ITI, an EEG screen approximately equal to the average background noise (−0.7879, 0.0000, 0.4667) was displayed. This procedure minimizes the contrast between the ITI background and the luminance noise background onset. The ITI screen also served to maintain the animals in a primarily photopic visual state because its luminance was relatively high (20.66 cd/m2). Fixation trials were run in a separate block from the MGS task. The three stimulus types were presented randomly interleaved with 20 presentations of each stimulus type. Stimuli were presented in the recorded neuron's receptive field at the calibrated tritan vector locations for each animal (Hall & Colby, 2013). Possible stimulus locations were located radially from fixation at an eccentricity of 4° for monkey FS and 6° for monkey CA.

Figure 1. 

Fixation task and average neuronal responses to background luminance noise and stimuli. Upper schematics show fixation task timing and paradigm. Center task schematic shows only the presentation of the high-contrast S-cone stimulus highlighted with a white outline, which is not present during the actual task. Neuronal responses are average spike density functions across all three stimulus types. Neurons were classified as sustained visual (SV), transient visual (TV), sustained visuomotor (SVM), and transient visuomotor (TVM). Left: During the ITI the screen is steady EEG. The luminance noise background is turned on, and at the same time a fixation cross appears. Middle: After 200–400 msec of fixation, a stimulus appears in the recorded neuron's receptive field. The stimulus was either a high-contrast S-cone, low-contrast S-cone, or luminance contrast. Right: The animal fixates for an additional 200–400 msec after stimulus presentation to receive a liquid reward.

Figure 1. 

Fixation task and average neuronal responses to background luminance noise and stimuli. Upper schematics show fixation task timing and paradigm. Center task schematic shows only the presentation of the high-contrast S-cone stimulus highlighted with a white outline, which is not present during the actual task. Neuronal responses are average spike density functions across all three stimulus types. Neurons were classified as sustained visual (SV), transient visual (TV), sustained visuomotor (SVM), and transient visuomotor (TVM). Left: During the ITI the screen is steady EEG. The luminance noise background is turned on, and at the same time a fixation cross appears. Middle: After 200–400 msec of fixation, a stimulus appears in the recorded neuron's receptive field. The stimulus was either a high-contrast S-cone, low-contrast S-cone, or luminance contrast. Right: The animal fixates for an additional 200–400 msec after stimulus presentation to receive a liquid reward.

Data Analysis

We analyzed trial event timing, eye position, and spike timing. All analyses were carried out off-line using custom MATLAB software (Mathworks, Natick, MA).

Neuron Classification

Neurons were classified by responses during the MGS task. Average neuronal firing rate during the baseline epoch (−100 to 0 msec before stimulus presentation) was compared with average activity during the visual response epoch (30–130 msec after stimulus presentation) and saccade epoch (−50 to 50 msec after saccade onset, defined as the time at which eye velocity first exceeded 30°/sec). Of 194 neurons recorded, 178 showed a significant visual response (one-sided t test, p < .05) and were included in further analysis. Of these neurons, 91 also had a significant saccade-related response (one-sided t test, p < .05) and were classified as visuomotor neurons. The remaining 87 neurons were classified as visual.

Visual and visuomotor neurons were further classified as having transient or sustained visual responses by computing a transience index. Previous studies have measured transience by comparing peak firing rate during a defined response epoch to average activity in the following epoch (White et al., 2009; Schiller & Malpeli, 1977). Our sample contained many cells exhibiting a strong response to stimulus offset as well as stimulus onset (e.g., Figure 2, Row 2). Neurons with clear on–off responses were generally transient in nature. Their on response must end before the off response begins to produce a distinct off response. Because of their off response, these cells have high levels of activity in the epoch following the initial response. Using previous approaches (White et al., 2009; Schiller & Malpeli, 1977), such neurons were often classified as sustained neurons, although their responses appeared very transient. We therefore used a different method for determining transience based on the Fourier transform of each neuron's spike trains. We binned the spike trains (0.1 msec time bins, i.e., 10,000 Hz sampling rate) of each neuron from −200 to 300 msec after stimulus presentation across all 40 trials of the MGS task and performed a fast Fourier transform on the average spike train. From the Fourier transform, we computed the average power at frequencies from 2 to 100 Hz as the baseline power. This frequency band was chosen because it contains spiking activity from neurons with high baseline firing rates (strong low frequency components) and neurons with low, spurious baseline firing rates (strong high frequency components). We then computed the average power at frequencies from 13 to 30 Hz as the stimulus power. This stimulus frequency band was chosen because it captures our stimulus frequency of 21.25 Hz (refresh rate divided by stimulus frames, i.e., 85/4) and encompasses the frequency power exhibited by neurons whose responses closely tracked stimulus presentation with on and/or off responses. Transience indices were then computed as the ratio: (stimulus power)/(stimulus power + baseline power). Thus, a neuron whose average baseline power was equal to its average stimulus power would have a transience index of 0.5. Neurons with transience indices of >0.5 were classified as transient and those of ≤0.5 were classified as sustained.

Figure 2. 

Rasters and histograms from example neurons in each of the four cell classes. The MGS task was used to classify neurons (left columns). Neuronal responses to the three stimuli were then recorded during the fixation task (right columns).

Figure 2. 

Rasters and histograms from example neurons in each of the four cell classes. The MGS task was used to classify neurons (left columns). Neuronal responses to the three stimuli were then recorded during the fixation task (right columns).

Neurons in the superficial layers of the SC have visual responses that are generally transient (White et al., 2009; Marrocco & Li, 1977; Schiller & Koerner, 1971). In this study, 20/31 visual neurons classified as transient were confirmed to lie near the collicular surface (less than 1 mm). Visual neurons deeper than 1 mm had sustained or transient responses, and at greater depths visuomotor neurons predominated.

Spike Density Function

Average spike density functions (SDFs) were computed by placing spikes across all trials from all neurons in 0.1-msec bins. The resulting nonnormalized histogram was then convolved with a Gaussian kernel using a 2-msec standard deviation and values were converted to firing rates in spikes per second.

Neuronal Response Latency

Neuronal response latency was determined by finding the time to half height of the peak response for each neuron (Lee, Williford, & Maunsell, 2007). To find the peak response, we computed the SDF for each neuron but using a 5-msec Gaussian kernel, which produces greater smoothing and reduces spurious results. We identified the time of SDF peak firing rate in the window from 40 to 150 msec after stimulus onset. The peak time was used to find the last point in time the SDF crossed threshold before reaching its response peak. Threshold was set at half the peak value plus the average SDF rate from 200 to 0 msec before stimulus onset. The time at which the firing rate last crossed threshold before reaching its peak was defined as the neuronal response latency. For some neurons, the response to one of the stimulus types was absent or very small, so that an accurate half height could not be determined in the prescribed window. We required that a response latency was determined for all three stimulus types to eliminate any bias caused by not including latency to all stimulus types from all neurons. Neurons whose response latency to all three stimulus types could not be determined were omitted from further latency analysis (15/56 SV, 8/31 TV, 12/66 SVM, and 7/25 TVM neurons were omitted).

Contrast Sensitivity Indices

For each neuron, we measured how its responses differed across the three stimulus types. To do this, we computed two contrast sensitivity indices: a high versus low S-cone contrast sensitivity index and a high S-cone versus luminance contrast sensitivity index. These indices were computed as (HighS − LowS)/(HighS + LowS) and (HighS − Luminance)/(HighS + Luminance) where HighS, LowS, and Luminance indicate the mean firing rate of the neuron to the respective stimulus types. The mean responses were computed from spikes in the interval 40–200 msec after stimulus presentation. These indices provide a measure of each neuron's change in response to the stimulus types compared. Values of zero indicate equal response and positive values indicate stronger response to the high-contrast S-cone stimulus.

Statistical Analyses

In the analyses on mean firing rates, we computed the mean firing rate for each neuron in the window from 40 to 200 msec after stimulus presentation. When using peak firing rates, we chose the same peak used for the latency analysis: maximum firing rate of the SDF for a single neuron in the window from 40 to 150 msec after stimulus onset. Peak firing rates were analyzed because, by definition, sustained neurons have longer responses than transient neurons. This could lead to a systematic bias toward sustained neurons exhibiting greater mean responses when comparisons are made across cell classes. We report analyses on peak firing rates primarily to make across class comparisons and include the same analyses on mean firing rates for comparison.

For comparisons across trials within a single neuron, we used a one-way ANOVA on mean firing rates of single trials. For comparisons of firing rate across neurons within a class, we found the mean response of each neuron across trials and used a one-way repeated-measures ANOVA to account for the firing rate differences across neurons. For comparisons of neuronal response latency across neurons within a class, we used the nonparametric Friedman test, which is comparable to a one-way repeated-measures ANOVA.

To compare population firing rates of all neurons across classes (SV, TV, SVM, and TVM), we performed a two-way ANOVA on both mean and peak firing rate with factors of Cell Class and Stimulus Type. This procedure allowed us to compare across all neurons in an unbalanced design and gives a measure of the interaction effect between cell class and stimulus type. The interaction statistic is used to test whether neurons in different classes responded differently to the three stimulus types. To compare neuronal response latency across classes, we wanted to continue using nonparametric tests and so chose to compare the latency across all four classes one stimulus type at a time. To this end, we used the Kruskal–Wallis test, which is comparable to a one-way ANOVA. All post hoc comparisons were made using Tukey's honestly significant difference test for multiple comparisons, denoted HSD. For correlation analyses, we used Pearson's correlation coefficient and the t statistic for significance testing.

RESULTS

We found that single neurons in macaque SC are activated by S-cone isolating stimuli. We measured the visual response characteristics of SC neurons to a luminance and two S-cone contrast stimuli, a high and low S-cone contrast. The prevailing hypothesis is that SC neurons cannot be activated by any S-cone isolating stimulus. We tested responses to two different S-cone contrasts because if SC neurons truly respond to S-cone contrast, their responses should change as S-cone contrast changes. We used a luminance noise background to eliminate luminance artifacts during a fixation task (Figure 1, upper schematics). Stimuli were presented at screen locations at which S-cone isolating stimuli had been calibrated in each animal. At the beginning of each trial, SC neurons responded weakly to the onset of the luminance noise background (Figure 1, lower plots). After the response to onset of the checkered background, neurons slowly adapted their firing rates, as shown previously for repeatedly stimulated SC neurons (Boehnke et al., 2011; Marrocco & Li, 1977; Cynader & Berman, 1972). After adaptation SC neurons exhibited sustained firing rates slightly above the baseline response during the ITI. Presentation of a visual stimulus embedded in the luminance noise elicited a large visual burst. In the sections below, we report how S-cone-specific contrasts affected single unit responses, average activity, peak activity, and neuronal response latency.

SC Neuronal Responses Modulate with S-cone Contrast

Single Neurons Respond to S-cone Stimuli

We divided SC neurons into four classes. First, we used the MGS task to classify neurons as visual or visuomotor. Of 194 neurons recorded, 178 showed significant visual responses during the MGS task. Of these, 91 showed significant saccade responses and were classified as visuomotor. The remaining 87 were classified as visual. Second, we classified neurons as either sustained or transient based on their visual response profile in the MGS task. Neurons from all four cell classes responded to the luminance and S-cone contrast stimuli.

Characteristic responses of neurons from each class are shown in Figure 2. The example sustained visual (SV) neuron (Figure 2, top row) responded to stimulus onset and had no response during the saccade in the MGS task. In the fixation task, this neuron fired less to the low-contrast S-cone stimulus than to the luminance and high contrast S-cone stimuli (one-way ANOVA, HSD, both p < .05). The transient visual (TV) neuron (Figure 2, second row) showed very brief responses to both stimulus onset and offset. Like the SV neuron, this TV neuron fired less to the low-contrast S-cone stimulus than the other two stimuli (p < .05). The example visuomotor (SVM and TVM) neurons (Figure 2, bottom 2 rows) had saccade related activity in the MGS task. The TVM neuron responded more briskly to stimulus presentation than did the SVM neuron. The example visuomotor neurons were sensitive to all three stimulus types but did not significantly distinguish among them (p > .05). Similar response patterns were observed across the population of SC neurons (Table 1). A total of 56/178 visual neurons responded significantly more to high than to low S-cone contrast stimuli. Most of the other neurons responded to all stimulus types, but response differences did not reach significance. The two S-cone stimuli were closely matched in their luminance contrast, differing only in their ability to excite S-cone sensitive visual inputs. These data show that most SC neurons respond to S-cone stimuli and that the activity of many SC neurons distinguishes S-cone contrast.

Table 1. 

Number of Neurons in Each Class that Discriminated between One or More of the Stimulus Types

Class
Total
Total Sig
High > Low
High > Lum
Lum > Low
SV 56 26 18 13 
TV 31 12 
SVM 66 23 18 13 
TVM 25 14 11 
Class
Total
Total Sig
High > Low
High > Lum
Lum > Low
SV 56 26 18 13 
TV 31 12 
SVM 66 23 18 13 
TVM 25 14 11 

High and Low refer respectively to the high-contrast and low-contrast S-cone stimulus responses. Lum refers to the luminance stimulus response. Counts in the Total column are the total number of neurons in each of the four classes. Counts in the Total Sig column are the number of neurons that showed significant response differences to the three stimulus types (one-way ANOVA, p < .05). The three rightmost columns show the number of neurons in each class with significant differences in mean response between the stimulus types indicated in the top row (post hoc, Tukey's HSD, p < .05). Neurons in the three rightmost columns are subsets of Total Sig but are not mutually exclusive.

All Cell Classes Are Sensitive to S-cone Contrast

We asked whether each class of SC neurons was able to discriminate the three stimulus types. The average SDF responses to each stimulus for neurons in the four classes are shown in Figure 3. All classes exhibited differential mean responses to the three stimulus types that were highly significant (repeated-measures one-way ANOVA, each p < .0001). Average firing rate for each class strongly and significantly differentiated between the high- and low-contrast S-cone stimuli (all classes, HSD, p < .001). In addition, all classes preferred the luminance stimulus to the low-contrast S-cone isolating stimulus (SV, TV, and SVM neurons, HSD, p < .001; TVM neurons, HSD, p < .05). We conclude that each class of neurons, taken as a whole, modulates their response with S-cone contrast.

Figure 3. 

Average SDFs for responses to the three stimulus types (luminance, black; high-contrast S-cone, dark blue; low-contrast S-cone, light blue). Sustained neurons are shown in the left column, and transient neurons are in the right column. Visual neurons are shown in the top row, and visuomotor neurons are in the bottom row. All SDFs are aligned on stimulus presentation.

Figure 3. 

Average SDFs for responses to the three stimulus types (luminance, black; high-contrast S-cone, dark blue; low-contrast S-cone, light blue). Sustained neurons are shown in the left column, and transient neurons are in the right column. Visual neurons are shown in the top row, and visuomotor neurons are in the bottom row. All SDFs are aligned on stimulus presentation.

Stimulus Response Modulation Is Similar across Cell Classes

We were especially interested in whether the different neuron classes responded differently to each stimulus type, as has been reported for luminance and color stimuli (White et al., 2009). To see how response modulation might differ across the four cell classes, we performed a two-way ANOVA on the mean neuronal responses using factors of Stimulus Type and Cell Class. To assess whether all cell classes were similarly modulated by the three stimulus types we looked at the interaction between Class and Stimulus. The result was highly nonsignificant (p = .999), indicating that the three stimuli were treated uniformly by all four cell classes.

We were concerned that using mean responses could affect our analysis of the interaction between class and stimulus type. The mean response of transient neurons is likely to be weaker than that of the sustained neurons in a large time window because sustained neurons were defined to have greater response duration. Indeed, the two-way ANOVA on average firing rates showed a strong main effect of Cell Class (p < .0001), which would be expected if sustained neurons fired more spikes to the stimuli than transient neurons. Post hoc comparisons revealed that SVM cells had stronger responses than both transient classes whereas SV neurons exceeded the response of TV neurons (HSD, all p < .05). Yet the average peak responses of each cell class are qualitatively congruent (Figure 3). White et al. (2009) reported a peak firing rate modulation of SC neuronal responses to colored stimuli. It is possible that many SC neurons distinguish the stimuli by modulation of their peak firing rate rather than mean rate.

We performed a second two-way ANOVA on the peak firing rates of the SC neurons to address this issue. The analysis showed that peak firing rate differences are invariant across cell classes (main effect of Class, p = .4950). We again saw that the four cell classes treated the three stimulus types similarly in their peak response modulation (interaction of class and stimulus, p = .9973). Thus, differential responses to the three stimuli over the population of SC neurons are independent of cell class.

SC Population Responses Differentiate All Three Stimuli

We wanted to know whether SC responses to S-cone stimuli could be greater than those to luminance at the population level. The two-way ANOVA on mean responses revealed a strong effect of Stimulus Type across all four classes of SC neurons (main effect of Stimulus, p = .0153). Post hoc analysis revealed that this discrimination was the same as for within class comparisons: Response to the high-contrast S-cone stimulus was significantly greater than the low (HSD, p < .05) but not the luminance stimuli (HSD, p > .05), whereas response to the luminance stimulus did not differ significantly from the low S-cone contrast (HSD, p > .05). The results were similar and even stronger when peak responses were analyzed (main effect of Stimulus, p = .0001). Post hoc analysis further confirmed that peak firing rate modulation follows the same pattern as mean firing rate modulation. Peak responses are greater to the high- than low-contrast S-cone stimulus (HSD, p < .001), whereas peak responses to the luminance versus low S-cone and luminance versus high S-cone were not significantly different (HSD, p > .05).

The population SDF response for each cell class is stronger for the high-contrast S-cone stimulus than the luminance stimulus (Figure 3). Yet these differences were not significant within any given class or in the two-way ANOVA pooling the SC population across classes. To determine whether these differences were significant, we performed an analysis that is not confounded by variance across neurons. Because neuronal responses across cell classes did not significantly differ in their modulation by stimulus type, we discarded the factor of class and performed a repeated-measures one-way ANOVA on the entire population of 178 SC neurons. In this analysis, we are not sorting by cell class, and the irrelevant variance across neurons is eliminated. The only effect under inquiry is that of stimulus type. We found that the high-contrast S-cone stimulus response was greater than the luminance, which in turn exceeded the low-contrast S-cone stimulus response (HSD, all p < .001). This pattern was identical and even more striking when peak activation was considered instead of mean response (HSD, all p < 1.0 × 10−9). Together these results show that the most important characteristics to SC neurons are the total contrast of stimuli, regardless of the types of cones excited. Responses scale with S-cone contrast and sufficiently high S-cone contrast stimuli can evoke stronger responses than a low-contrast luminance stimulus.

SC Neuronal Response Latency Modulates with S-cone Contrast

All Classes Shift Neuronal Response Latency with S-cone Contrast

It is typical of visual neurons to modulate both response magnitude and latency as a function of contrast. As contrast increases, latency decreases. If SC neurons are truly sensitive to S-cone contrast, their response latency should change as S-cone isolating contrast changes. We plotted cumulative distributions of neuronal response latencies for each class of neuron and found that this was indeed the case (Figure 4). Within each class, response latency was significantly modulated across the three stimulus types (Friedman test, all classes p < 1.0 × 10−6). Neuronal response latencies to high-contrast S-cone and luminance stimuli were much shorter than to the low-contrast S-cone for all classes (HSD, both p < .01). In addition, the TV neurons had significantly shorter latency responses to the high-contrast S-cone than to the luminance stimulus (HSD, p < .05). These results mirror those found with neuronal activity, namely, that responses to the high-contrast S-cone stimulus significantly differed from those to the low-contrast S-cone stimulus within each cell class.

Figure 4. 

Cumulative distributions of neuronal response latency. Conventions are the same as in Figure 3. Colored numbers and vertical lines indicate the median response latency to each corresponding stimulus type. Neurons are drawn from the same population and cell classes as in Figure 3. Neurons were excluded if the response to at least one stimulus type was too weak for latency analysis (see Data Analysis in Methods).

Figure 4. 

Cumulative distributions of neuronal response latency. Conventions are the same as in Figure 3. Colored numbers and vertical lines indicate the median response latency to each corresponding stimulus type. Neurons are drawn from the same population and cell classes as in Figure 3. Neurons were excluded if the response to at least one stimulus type was too weak for latency analysis (see Data Analysis in Methods).

Neuronal Response Latency Shifts Are Greater for Visual Neurons

Given that all classes had response latencies modulated by S-cone contrast, we wanted to know whether there were any differences across the neuron classes. To that end, we compared the neuronal latency distributions to each stimulus type across the four cell classes. We found that neuronal latency to the luminance and low-contrast S-cone stimuli were roughly the same across classes (Kruskal–Wallis, p = .0563 and p = .1759, respectively). For the high S-cone contrast stimulus, neuronal latencies differed significantly (Kruskal–Wallis, p = .0169), with SV neurons responding earlier than SVM neurons (HSD, p < .05). Although this comparison yielded the only significant difference, neuronal latency differences to the luminance stimulus across cell classes approached significance. It is likely that our sample of transient neurons was inadequate to detect any small differences in latency between TV and visuomotor neurons. Furthermore, because stimuli that elicit the largest visual responses (the high S-cone and luminance stimulus) produced the greatest differences across classes, nonlinearities in contrast sensitivity between transient and sustained neurons may have obscured differences between neuronal classes. Qualitatively, the median latencies of the TV neurons were all less than or equal to those of the TVM neurons. These data suggest that visual neurons may respond sooner than visuomotor neurons to high-contrast stimuli, although the differences appear small.

SC Population Neuronal Response Latencies Differentiate All Three Stimuli

Finally, we wondered whether the small differences seen in median response latency between the luminance and high-contrast S-cone stimuli were meaningful at the population level, independent of cell class. To increase statistical power, we grouped response latencies across all cell classes, despite the possibility of small differences in neuronal response latency between visual and visuomotor neurons. This analysis yielded a highly significant result for latency differences across the stimulus types (Friedman, p < 1 × 10−33). Further analysis confirmed that response latencies to each of the three stimulus types significantly differed from one another (HSD, all p < .001). In particular, the neuronal response latencies to the high-contrast S-cone stimulus were shorter than to the luminance stimulus. SC neurons within each class responded with shorter latency to high than to low S-cone contrasts. At the class-independent population level, increasing S-cone contrast elicited responses with shorter latency compared with the luminance contrast stimulus.

S-cone Contrast Sensitivity Increases with Transience

Throughout the visual system, there is a tendency for TV neurons to be more sensitive to changes in contrast. We asked whether this would be true of SC neurons using S-cone isolating contrasts. We plotted the high versus low S-cone contrast sensitivity index of each neuron against its transience index separately for visual and visuomotor neurons (Figure 5, top row). Nearly all neurons tested had a stronger mean response to high- than to low S-cone contrasts (73/87 visual and 82/90 visuomotor contrast sensitivity indices > 0). This shows that although only about 1/3 of neurons significantly distinguished the two contrasts (Table 1), nearly all neurons increased their response as S-cone contrast increased. In addition, the degree to which neurons modulated their response with S-cone contrast was correlated with their level of transience (visual cells, r = 0.2121, p = .0486; visuomotor cells, r = 0.2885, p = .0058). SC neurons with more transient responses were more sensitive to changes in S-cone contrast.

Figure 5. 

Contrast sensitivity indices as a function of transience index. Top row: High versus low S-cone contrast sensitivity index. Each point plots a neuron's contrast sensitivity index against its transience index. Colors indicate each neuron's classification as visual or visuomotor. Filled circles indicate neurons classified as sustained, and open circles indicate neurons classified as transient. Solid black lines are the least squares linear regression line of best fit. Dashed horizontal black lines correspond to a contrast sensitivity index of 0, indicating equal responses to high- and low-contrast S-cone stimuli. Bottom row: High S-cone versus luminance contrast sensitivity index. Conventions are the same as top row.

Figure 5. 

Contrast sensitivity indices as a function of transience index. Top row: High versus low S-cone contrast sensitivity index. Each point plots a neuron's contrast sensitivity index against its transience index. Colors indicate each neuron's classification as visual or visuomotor. Filled circles indicate neurons classified as sustained, and open circles indicate neurons classified as transient. Solid black lines are the least squares linear regression line of best fit. Dashed horizontal black lines correspond to a contrast sensitivity index of 0, indicating equal responses to high- and low-contrast S-cone stimuli. Bottom row: High S-cone versus luminance contrast sensitivity index. Conventions are the same as top row.

S-cone Contrast Sensitivity Is Weaker than Luminance

In addition to response modulations with S-cone contrast, we asked how the high-contrast S-cone stimulus responses compared with those to the luminance stimulus. We examined this question by plotting the high S-cone versus luminance contrast sensitivity index of each neuron against its transience index separately for visual and visuomotor neurons (Figure 5, bottom row). These data indicate that SC neuronal responses were much more similar for the high S-cone and luminance stimuli than between the two S-cone stimuli (45/87 visual and 55/91 visuomotor contrast sensitivity indices > 0). Correlations for visual and visuomotor neurons in this instance were not significant (visual cells, r = 0.2001, p = .0631; visuomotor cells, r = 0.0732, p = .4907), although visual neurons trend in the positive direction. The contrast of the high S-cone stimulus was much greater than the luminance in DKL space. The contrast of the low S-cone stimulus was only slightly less than the luminance stimulus contrast. Our results indicate that SC neurons as a whole treated the high S-cone and luminance stimuli as being more similar than the low S-cone and luminance stimuli, in opposition to their definition in DKL space. These data suggest that overall SC neurons may be more sensitive to luminance contrast than to S-cone contrast. Nevertheless, response magnitude can be made greater or smaller to an S-cone compared with a luminance stimulus by appropriately manipulating contrast.

DISCUSSION

Neurons in the SC are sensitive to S-cone isolating stimulus contrasts. SC visual responses are a function of total cone contrast that does not discriminate against S-cones. This finding raises questions about the visual properties of and visual inputs to SC neurons. It has a direct bearing on studies that have used S-cone isolating stimuli to probe the role of the SC in visual and oculomotor phenomena, such as blindsight.

How Could Information about S-cone Stimuli Reach the SC?

There are two major pathways by which visual input from the retina reaches neurons in the SC. The SC receives direct input from retinal ganglion cells, primarily to the superficial layers (Marrocco & Li, 1977; Bunt, Hendrickson, Lund, Lund, & Fuchs, 1975; Hubel, LeVay, & Wiesel, 1975; Hendrickson, Wilson, & Toyne, 1970). It also receives indirect, cortical input from striate and extrastriate cortex. Striate cortex targets more superficial neurons, whereas extrastriate cortex targets the intermediate and deeper layers (Sommer & Wurtz, 2004; Fries, 1984; Kuypers & Lawrence, 1967). Either or both of these pathways could contribute to the collicular activation we observed using S-cone stimuli.

Early Studies of SC Visual Afferents and Properties

In early studies, the ganglion cells comprising the direct, retinotectal pathway were reported to lack chromatic sensitivity. Pioneering physiological investigations of the direct pathway to SC recorded retinal ganglion cells, studied their properties, and confirmed tectal projection by antidromic activation from the SC (De Monasterio, 1978a, 1978b; Schiller & Malpeli, 1977). They found that direct projections to SC arose primarily from nonopponent ganglion cells that sum input from L- and M-cones. SC-projecting ganglion cells had transient responses and generally lacked a strong center-surround organization. Many of these ganglion cells were classified as the broad band type now associated with the magnocellular “luminance” pathway. The conclusion was that the direct pathway to the SC carries primarily or exclusively luminance contrast information from L- and M-cones.

Cortical projections to the SC also appear to stem from luminance channels in the LGN (Finlay, Schiller, & Volman, 1976). Antidromically activated corticotectal neurons in V1 are mostly complex cells that have large receptive fields, weak orientation selectivity and lack color opponency. Lesion or cooling of striate cortex causes a powerful, selective decrease in visual sensitivity of neurons in deeper SC layers while sparing sensitivity of the superficial layers (Schiller, Stryker, Cynader, & Berman, 1974). Similar deficits were found after inactivation of the magnocellular layers of the LGN, but not the parvocellular layers, where color opponent neurons are found (Schiller et al., 1979). The conclusion was that visual inputs to the SC from striate cortex arise from the magnocellular layers of the LGN, summing achromatic signals from L- and M-cones in the retina.

The properties of neurons projecting to the SC through both the direct and indirect pathways are reflected in those of visual SC neurons (Marrocco & Li, 1977; Cynader & Berman, 1972; Goldberg & Wurtz, 1972). Visual neurons in the SC have comparatively large receptive fields with a weak or absent inhibitory surround. They are relatively insensitive to stimulus size, shape, orientation and lack color opponency. Observations of SC neurons and their primary visual inputs have led to the belief that SC neurons do not get input from S-cones.

“Rarely Encountered Cells”

Initial reports on both direct and indirect pathways showed a lack of color opponent input with no contribution from S-cones. From an evolutionary perspective, the lack of S-cone input to the SC is a curious finding. Both the SC (Kaas, 2004) and S-cones are evolutionarily ancient (Mollon, 1989). The SC shares similarities across many mammalian species, and the ganglion cell types that carry S-cone signals in modern primates were more prominent in early primates (Kaas, 2004). Genes encoding the S-cone in macaques and humans are highly homologous and conserved across mammals, although not all mammals express this gene (Yokoyama & Yokoyama, 1989; Nathans, Thomas, & Hogness, 1986). Across new and old world monkeys, systems for opponent S-cone vision appear conserved, again hinting at its early origins among primates (Jacobs, 2007; Silveira et al., 1999), with the notable exception of some new world species (Levenson, Fernandez-duque, Evans, & Jacobs, 2007; Jacobs, 1998). The conservation of S-cones and the SC in mammals would appear to make their interaction likely.

Why have S-cone inputs to the SC gone undetected? The most likely explanation stems from the properties of the S-cone subsystem. S-cones, and the ganglion cells that carry their output are rare in the retina, comprising only about 5–10% of cells (Calkins, 2001; Bumsted & Hendrickson, 1999). Ganglion cells that project to the SC are also scarce and tend to have broad, sparse dendritic fields (Rodieck & Watanabe, 1993; Perry & Cowey, 1984; Leventhal, Rodieck, & Dreher, 1981). To make matters worse, ganglion cells carrying S-cone signals are a heterogeneous population that tend to have small axons and low firing rates (Hendry & Reid, 2000). As noted by Schiller and Malpeli (1977), these properties are likely to create a severe bias against detecting such ganglion cells in nontargeted extracellular recordings. Nevertheless, these researchers reported a population of “rarely encountered cells” with small axons and slow conduction velocity that projected to the SC in unusually high proportions. Another early report on retinal ganglion cells that project to SC called them “atypical,” with on and off responses to visual stimuli and nonconcentric receptive fields (De Monasterio, 1978b). These cells seem to correspond well with earlier reports of less numerous nonconcentric ganglion cells with phasic responses, large receptive fields, and no color opponency (De Monasterio & Gouras, 1975). Although not color opponent, some of these cells responded well to stimuli of any color, as do visual neurons in the SC (White et al., 2009; Marrocco & Li, 1977). Some atypical ganglion cells were reported to receive input from all cone types and even project directly to the SC (De Monasterio, 1978a). More recent studies have confirmed these findings, revealing a heterogeneous population of numerous ganglion cell types that project to the SC (Rodieck & Watanabe, 1993; Perry & Cowey, 1984; Leventhal et al., 1981). Many properties of SC projecting ganglion cells closely resemble those of ganglion cells known to carry S-cone-related signals to the koniocellular layers of the LGN (Dacey & Packer, 2003; Hendry & Reid, 2000). Chromatic sensitivity in the SC could arise from this small population of ganglion cells that have proven especially difficult to characterize.

Recent Characterization of S-cone Input to Retinal Ganglion Cells

Advances in technique have allowed targeted, intracellular recordings of retinal ganglion cells. Targeted recording led to the identification of the first class of ganglion cell that carries S-cone-specific signals (Dacey, 1996; Dacey & Lee, 1994). Modern tracing techniques allow morphological identification of recorded neurons. Targeted recording combined with chromatic adaption and silent substitution (older studies used monochromatic lights to study color opponency, not cone isolation techniques) allows physiological characterization of cone inputs in addition to anatomical imaging (Dacey, 1999). Without these methods, it is difficult to measure cone-specific input, especially if S-cone input is weak, and nearly impossible to classify ganglion cell types (Dacey, Peterson, Robinson, & Gamlin, 2003; Klug, Herr, Ngo, Sterling, & Schein, 2003; Calkins, 2001). It is now known that a diverse population of ganglion cells carries both S-cone on and off inputs to the more recently recognized koniocellular layers of the LGN, which lie between the magnocellular and parvocellular layers (Szmajda, Buzas, FitzGibbon, & Martin, 2006; Hendry & Reid, 2000; Martin, White, Goodchild, Wilder, & Sefton, 1997). Most SC projecting ganglion cells have been classified as the P gamma cell type (Perry & Cowey, 1984). This heterogeneous population is most commonly associated with the diverse cell types that receive S-cone input and project to the koniocellular layers of the LGN. Whether the cells so far classified as receiving S-cone input also project to the SC is not yet known.

There are a large number of cell types and cell-specific interactions in the retina whose properties are still being characterized (Dacey & Packer, 2003; Dacey, 1999). Efforts to study these interactions have revealed a variety of sparse cell types that carry input from S-cones and are similar in many respects to those known to project to the SC. Additional studies suggest that S-cones may influence other visual subsystems more than originally believed. For example, S-cones contribute to a number of functions commonly associated with the luminance channel, and it has been suggested that they may contribute to this pathway (Calkins, 2001). S-cone OFF bipolar cells have been found that contact midget ganglion cells and hence may influence the parvocellular pathway (Klug et al., 2003). Other research has uncovered bipolar cells that could contact all cone types (Joo, Peterson, Haun, & Dacey, 2011) and melanopsin-expressing ganglion cells that carry color signals derived from S-cones and may contribute to perception (Dacey et al., 2005). At least eight other types of ganglion cells were recently discovered, representing only 1–2% of the population, and at least two of these are sensitive to S-cone-specific excitation (Dacey & Packer, 2003; Dacey et al., 2003). In addition, two more types of achromatic ganglion cells have been reported that sum input from all cone types (Calkins & Sterling, 2007). Many of these cells have wide dendritic fields, suggesting broad summation and large receptive fields, as would be expected for SC projecting cells and is common among koniocellular projecting cells (Szmajda, Grünert, & Martin, 2008). Many other types of small ganglion cells have been observed but await further classification to determine their cone sensitivity and possible tectal projections (Crook, Peterson, Packer, Robinson, Gamlin, et al., 2008; Crook, Peterson, Packer, Robinson, Troy, et al., 2008). Ongoing research makes it clear that the types of cells and signals leaving the retina are not fully understood. Some of these newly discovered ganglion cell types or those yet to be fully characterized could be the source of S-cone input observed in the SC.

Comparisons to Previous Results on S-cone Related Excitation in the SC

Several previous studies have examined the effect of color and S-cone stimuli on visual responses in the SC. These studies have used a variety of techniques. None have used S-cone stimuli calibrated individually for each observer and spatial location combined with dynamic luminance noise. Our focus was on using techniques from human psychophysics to make our results directly comparable.

Chromatic Sensitivity in Macaque SC

Only one previous study has investigated color sensitivity in SC neurons recorded in awake, behaving macaques (White et al., 2009). Isoluminant Gaussian patches were presented in the receptive field of SC neurons. The color of the patches was chosen from across DKL space, including stimuli near and along the theoretical, uncalibrated, S-cone isolating direction. Recorded SC neurons were broadly sensitive to color, including colors that should most strongly modulate the S-cone opponent mechanism. White et al. (2009) further reported that achromatic stimuli produce shorter latency visual responses than chromatically defined stimuli. This latency result may reflect the use of only a maximum contrast luminance stimulus. We used a low-contrast luminance stimulus and varied the contrast of our chromatic S-cone stimulus. Our data show that, although low-contrast S-cone stimuli evoke longer latency responses than luminance stimuli, a high-contrast S-cone stimulus can evoke shorter latency responses than a luminance stimulus. When comparing chromatic versus achromatic response latency, stimulus contrast must be taken into account. Our results indicate that neuronal responses in the SC are largely determined by a contrast-dependent broadband summation of cone activation that includes S-cones.

Sustained and transient SC neurons differ in their sensitivity to color as compared with luminance (White et al., 2009). White et al. (2009) found that TV neurons are much less sensitive to color than to luminance. We performed a similar neuronal classification on our visual SC neurons. We found that the most transient neurons are more sensitive to changes in S-cone isolating contrast. Differences between the high and low-contrast S-cone stimuli in the present results and the findings between the luminance and chromatic stimuli of White et al. (2009) are similar on the grounds of contrast. In both studies, TV neurons were more sensitive to contrast. Our data show that contrast sensitivity is not specific to the particular visual mechanism activated by the stimulus. We conclude that TV neurons are more sensitive to contrast than sustained neurons and this includes chromatic sensitivity to S-cone input.

Chromatic Sensitivity in Marmoset SC

Interesting recent investigations in new world monkeys measured cone input to SC neurons in anesthetized marmosets, with emphasis on input from S-cones (Tailby et al., 2012). S-cone responses were observed only after presentation of flashed S-cone stimuli. These responses were typically weak, although the S-cone response was stronger to stimulus offset. The relatively weak response of SC neurons to S-cone stimuli was attributed to residual contrast in long wavelength cones and not true S-cone input. When drifting gratings were used to measure response characteristics of SC neurons, responses to changes in S-cone contrast were not observed. The authors concluded that SC neurons are not driven by S-cones.

Residual contrast in longer wavelength cones seems insufficient to account for our data. The SC neurons we recorded frequently exhibit on and off responses to S-cone stimuli. These responses are as large as those to stimuli designed to specifically modulate contrast in the long wavelength sensitive cones. We used two levels of S-cone contrast and, across animals and spatial locations, presented six radiometrically distinct S-cone stimuli. The effects of these stimuli on the luminance and L-M opponent channels were small and, because of limitations of the 8-bit DAC, inconsistent across stimuli. S-cone stimuli were presented on a checkered background that was constantly changing luminance to mask residual responses in luminance channels by deliberately creating luminance contrast artifacts (Birch et al., 1992). Because the monitor calibration is not perfect and equiluminant points vary between neurons (Schiller & Colby, 1983), the noise had small additional effects on L-M opponent channels. Yet we observed consistent and large SC neuronal response modulation with changes in S-cone contrast. Considering that our use of luminance noise is, by design, better suited to control for residual luminance effects than L-M opponent effects, it is worth noting what would be expected of residual L-M opponent activation. The L-M opponent channel is slower than the luminance channel, both psychophysically and electrophysiologically (Bompas & Sumner, 2008; Smithson & Mollon, 2004; Schmolesky et al., 1998; Maunsell & Gibson, 1992). This suggests that it would take a large amount of residual L-M opponent contrast for SC neurons to exhibit shorter response latencies to the high-contrast S-cone stimulus than to the luminance stimulus. Taken together with the relatively large response magnitude to chromatic stimuli observed in our data and in that of White et al. (2009), it is difficult to attribute our responses entirely to residual activation of L- and M-cones.

The conclusion that S-cone input does not reach the SC is surprising when one considers the other physiological and anatomical aspects of this structure. In addition to their analysis of cone inputs, Tailby et al. (2012) rigorously characterized the temporal, spatial, and direction preferences of SC neurons in a way not previously done. They found that spatiotemporal properties of SC neurons more closely match those of koniocellular layer neurons, which are known to receive input from S-cone-sensitive ganglion cells, than those of the magnocellular or parvocellular layers, which do not. These data in isolation would predict the presence of S-cone input to the SC if neuronal responses are largely a reflection of the properties of their afferent visual neurons.

Two major differences in experimental procedure compared with the awake, behaving macaque model were noted by Tailby et al. (2012). (1) Their subjects were anesthetized and the effects of anesthesia are prejudiced against cortical structures. Such an effect could disproportionately reduce cortically dependent color sensitivity in the SC. (2) Despite their many similarities to macaques, it is possible that new world marmoset monkeys have organizational differences in their visual system. This possibility is emphasized by discrepancies between new world and old world monkeys in their S-cone and koniocellular layer organization (Jacobs, 2007; Hendry & Reid, 2000). Especially of interest here is that the projection from SC to koniocellular LGN neurons appears stronger in macaques than in new world monkeys, suggesting possible differences for the role of S-cone input and tectal influence between these species (Hendry & Reid, 2000). The observation that S-cone input was not present in the marmoset SC could be a result of either of these factors.

Chromatic Sensitivity in Human SC

How are physiological results in monkeys reflected in human fMRI responses? This question has been directly addressed in two recent studies. In both studies, S-cone isolating stimuli were presented while measuring fMRI activation in the SC. Leh et al. (2010) compared activation to achromatic and S-cone stimuli where the achromatic stimulus was of greater cone contrast. They found that activation of the SC was absent using S-cone compared with achromatic stimuli (Leh et al., 2010). Interestingly, a second fMRI study appears to have measured a BOLD signal response change to S-cone stimuli in the same direction as luminance stimuli during pro and antisaccade tasks (Anderson, Husain, & Sumner, 2008). Their effect suggests stronger activation to luminance than to S-cone stimuli that were matched for subjective salience. The conclusion from both studies was that the human SC is not activated by S-cone isolating stimuli.

Three major concerns arise when considering fMRI activation of the SC to visual stimuli. (1) The SC is a difficult structure to measure using fMRI (Anderson & Rees, 2011), although rough topography and visual responses have been measured (Schneider & Kastner, 2005; DuBois & Cohen, 2000). (2) Visual responses in the SC rapidly habituate (Tailby et al., 2012; Boehnke et al., 2011), which is likely to reduce measured visual responses given the temporal resolution of fMRI. (3) The final and most important concern is stimulus contrast. Our data and those from other studies (Bell, Meredith, Van Opstal, & Muñoz, 2006; Schneider & Kastner, 2005) demonstrate that contrast has a major effect on SC activation. Even when cone contrast is matched, it is difficult to make comparisons between cone mechanisms (Brainard, 1996), particularly because the luminance mechanism shows a steeper, saturating CSF (Tailby et al., 2012; Tailby, Szmajda, Buzas, Lee, & Martin, 2008). Data from these two fMRI studies rely on changes relative to activation during a luminous background presentation and not absolute activity. This would make it easy to miss responses in the SC to S-cone stimuli that are weaker than responses to other stimuli. Indeed, it appears Leh et al. (2010) observed diminished responses to the S-cone stimulus in several visual areas, and both studies show weak activation of the SC, which could be accounted for by differences in contrast.

In summary, direct tests of input to the SC show a number of differences likely to stem from the use of different species as well as stimulus presentation and measurement techniques. Whether there are actually large differences between species and stimulus presentation techniques requires further research. We chose to follow closely techniques used in human psychophysics to relate our data to this line of work. Unlike previous studies in nonhuman primates, we calibrated S-cone stimuli for each animal and spatial location and used a relatively strong luminance noise background to reduce residual response artifacts. Our technique revealed strong modulation of SC neurons with changes in S-cone contrast that contradict several other findings.

Blocking Visual Input to the SC with S-cone Stimuli

Beginning with the innovative work of Sumner and colleagues (2002), many human psychophysical studies have taken advantage of an apparent lack of S-cone input to explore the role of SC. The rationale for these experiments is that the SC can be effectively “lesioned” by using visual stimuli that the SC cannot perceive. Collicular “lesions” are performed by presenting stimuli that are only detectable through use of retinal S-cones. Behavioral responses to S-cone stimuli are then compared with those to luminance stimuli to which the SC is not blind. If differences (typically RT differences) are observed between behavioral responses to the S-cone and luminance stimuli, the conclusion is that the SC mediates the behavior under investigation because the SC cannot resolve the S-cone stimulus. A negative result using S-cone stimuli (RTs to luminance and S-cone stimuli are equal) is taken to mean that the SC does not mediate the behavioral effect in question. Our data show that the assumption that the SC is insensitive to stimuli that activate only S-cones is incorrect. Nevertheless, several studies have reported behavioral effects using S-cone stimuli.

The first experiments to use S-cone stimuli to block the SC reported distinct behavioral effects for luminance compared with S-cone stimuli. The first study to use S-cone stimuli to block visual input to SC investigated two phenomena: the oculomotor distractor effect and exogeneous orienting of attention (Sumner et al., 2002). The oculomotor distractor effect is characterized by increases in saccadic RT to a visual target when a second visual stimulus, the distractor, is presented elsewhere in the visual field. Sumner et al. (2002) reported the absence of the oculomotor distractor effect when the distractors were S-cone isolating stimuli. This result was found only for saccadic responses and not when manual responses were required. Exogenous orienting of attention is the phenomenon whereby an irrelevant stimulus briefly presented at the target location just before the go cue decreases RT. In contrast to the distractor effect, S-cone and luminance cues both produced exogenous orienting of attention. The conclusion was that the SC has a primary role in the oculomotor distractor effect when saccadic responses are required but not in exogenous cueing of attention.

The initial success in finding distinct outcomes with S-cone compared with luminance stimuli prompted further studies. Inhibition of return is the tendency to avoid returning to previously attended locations. Inhibition of return was absent when saccadic responses were required to a previously attended S-cone stimulus location (Sumner, Nachev, Vora, Husain, & Kennard, 2004). In contrast, inhibition of return was observed after attending an S-cone stimulus when manual responses were required. This distinction suggested the possibility of two distinct mechanisms for inhibition of return: a saccadic mechanism mediated through the SC and a separate mechanism for manual responses independent of the SC (Sumner, 2006).

Continued use of S-cone isolating stimuli to block SC input showed no effect of using S-cone isolating stimuli on the gap effect and nasotemporal asymmetry (Bompas, Sterling, Rafal, & Sumner, 2008; Sumner, Nachev, Castor-Perry, Isenman, & Kennard, 2006). The gap effect is seen when a fixation target is extinguished before target presentation, providing a temporal gap between fixation release and motor response. The result is faster saccadic RT. Studies of the gap effect using S-cone stimuli as the fixation point were similar to those using a luminance defined fixation point (Sumner et al., 2006). Although fixation of the luminance stimulus provided a stronger effect, it could not be ruled out that the gap effect was sensitive to S-cone stimulus fixation. Nasotemporal asymmetry is seen when observers, monocularly presented with two stimuli, preferentially choose the target in the temporal visual field. This effect is thought to reflect an asymmetry in retinotectal projections. Nasotemporal asymmetry was found to be similarly present using both S-cone and luminance defined targets (Bompas et al., 2008). Results from these two studies suggested that the SC does not play a pivotal role in the gap effect or nasotemporal asymmetry.

The idea that visual input to the SC could be blocked or altered using S-cone stimuli, as cleverly set forth by Sumner and colleagues, proved highly influential. Numerous other groups have used S-cone stimuli to investigate SC function under the assumption that the SC is insensitive to S-cone stimuli. Phenomena studied range from the redundant target effect (Leo, Bertini, di Pellegrino, & La davas, 2008) and pro-antisaccade cost (Anderson et al., 2008) to more clinical applications such as interhemispheric transfer in callosotomy patients (Savazzi et al., 2007; Savazzi & Marzi, 2004) and mediation of blindsight (discussed below), among many others.

How Can Psychophysical Work Be Reconciled with the Current Results?

Our experimental methods were modeled after human psychophysics studies using S-cone stimuli with the goal of testing the assumption that the SC is blind to S-cone stimuli. Yet our data show that SC neurons are responsive to S-cone isolating stimuli under these conditions. This means that the SC cannot be blocked by using an S-cone stimulus in this manner.

How could psychophysical studies find positive results if S-cone activation does reach the SC? This is an important question given the widespread use of S-cone stimuli now present in the literature. Many of these studies did not use precisely measured or individually calibrated S-cone stimuli. They simply used “blue” or “purple” stimuli whose retinal activation properties are not known. Several also used low levels of luminance noise, which could lead to influence from rods. If we focus on those that closely follow the techniques proposed by Sumner and colleagues, there are two explanations. Both hinge on the fact that nearly all psychophysical studies using the S-cone stimulus technique rely heavily on RT measurements.

Stimulus Contrast Affects RT and Neuronal Response

It has long been known that RT is strongly dependent on stimulus contrast or magnitude (Hovland & Bradshaw, 1937). The response latency of SC neurons is largely determined by the luminance contrast of the stimulus (Bell, et al., 2006). Our data extend this finding to include S-cone isolating contrasts. Therefore, if S-cone and luminance stimuli are not matched, RTs could vary as a simple matter of contrast and neuronal latency.

It is noteworthy that several studies found negative results using S-cone stimuli when manual responses were required, but positive effects using saccadic responses. Saccadic eye movements are executed with shorter RT and greater velocity than typical manual responses. Our data demonstrate that neuronal latencies in the SC (as in other brain regions) are influenced by cone contrast. Furthermore, our data suggest that luminance signals are likely to reach the SC faster than S-cone signals of comparable strength. These small differences in transmission time would be more likely to affect motor systems with simple, rapid processing like the saccadic eye movement system. Thus, it is possible that manual and saccadic RTs are differentially influenced by S-cone stimuli because of differences in timing precision, rather than their explicit use of the SC visual pathways.

Supporting the idea that contrast-dependent neuronal response latency plays a critical role, the gap effect and nasotemporal asymmetry showed no effect of S-cone stimuli. In both tasks, the impact of differential processing time between the luminance and S-cone stimulus is likely to be negligible. For the gap task, the fixation point is turned off, and the subject has 200 msec to release fixation and prepare for the saccade. As long as this process takes less time than is actually allotted, differences in S-cone and luminance processing time would not be reflected in behavioral RT. The investigation of nasotemporal asymmetry actually avoided confounds of direct comparison between luminance and S-cone stimuli entirely. Using this paradigm, saccade bias was measured through comparisons within stimulus type. No behavioral differences were observed between S-cone and luminance stimuli when the differential processing time of luminance and S-cone stimuli did not affect the results. This further supports the conclusion that positive behavioral effects may have underlying causes that stem from neuronal response latency and stimulus contrast, rather than the role of SC in the behavior.

S-cone and Luminance Processing Channels Differ

The S-cone processing system appears slower than the luminance system. Several studies using S-cone stimuli have attempted to account for confounds of stimulus contrast by equating stimuli for subjective salience or objective contrast (Thirkettle et al., 2013; Anderson et al., 2008; Bompas & Sumner, 2008; Sumner et al., 2006). Nevertheless, differences in RT are often observed between subjectively matched S-cone and luminance stimuli. This could be attributed to differences in the S-cone opponent and luminance channels in the brain. Even subjectively matched luminance and S-cone stimuli vary in processing time psychophysically (Bompas & Sumner, 2008; Smithson & Mollon, 2004). Physiologically, differences are also observed in S-cone signal processing in V1 (Cottaris & De Valois, 1998) and through retinal ganglion cells and koniocellular neurons in the LGN (Hendry & Reid, 2000). Estimates of these differences typically indicate that S-cone processing is about 20–40 msec slower than luminance processing, which could have a substantial impact on measured RT.

Some investigators using S-cone stimuli have acknowledged the differential processing of S-cone and luminance signals. Attempts to account for the slowness of the S-cone system have involved manipulating cue timing and stimulus onset asynchronies to allow more time for S-cone signal processing (Bompas & Sumner, 2009; Sumner et al., 2004). Indeed, this technique has revealed that, contrary to Sumner et al. (2002), the oculomotor distractor effect is present when S-cone stimuli are used (Bompas & Sumner, 2009). This finding validates the importance of RT effects caused by stimulus contrast and processing in separate visual channels. The combination of subjective salience matching and stimulus onset asynchronies for S-cone stimuli appears to be a necessary and valuable technique for comparing RT between S-cone and luminance stimuli.

How Can Psychophysical Results Be Reinterpreted?

Our finding that S-cone stimuli activate SC neurons requires a reevaluation of results using such stimuli to block the SC. Despite the concerns raised above, there are positive results demonstrating differences in the use of S-cone and luminance contrasts that are likely attributable to the use of S-cone stimuli. Instead of these psychophysical findings being directly attributable to the SC per se, they are likely a reflection of the S-cone processing system itself. Three simple lines of evidence support the view that the S-cone and luminance systems should process stimuli differently. (1) The visual system can be subdivided into multiple distinct subsystems, where luminance and color are processed separately (Nassi & Callaway, 2009; Derrington, 2002). (2) Even the individual chromatic channels have evolved separately for their own purposes, being distinct anatomically, morphologically, and immunologically (Smithson & Mollon, 2004). (3) Most importantly, the S opponent system is known to exhibit a number of properties different from luminance and L-M opponent channels (Nassi & Callaway, 2009; Calkins, 2001; Hendry & Reid, 2000). On these grounds, behavioral and neuronal response differences between S-cone and luminance stimuli should be expected independent of collicular mediation. Because our data show that S-cone stimuli are able to drive SC neurons, previous results might be better interpreted in light of the peculiarities of the S opponent subsystem as a whole, rather than as an indicator of SC contribution.

Implications for the Study of Blindsight

SC and Extrastriate Cortex Play a Role in Blindsight

Blindsight is defined as the ability to detect, localize, and discriminate visual stimuli despite visual field defects produced by damage to primary visual cortex (V1). The term is used to acknowledge the fact that patients with V1 lesions may have quite good residual visual capabilities despite their inability to report visual detection verbally (Cowey, 2010; Weiskrantz, 2004; Sanders et al., 1974). Consistent with a lack of conscious awareness of visual stimuli, blindsight is distinctly different from impaired normal vision, such as near threshold detection (Azzopardi & Cowey, 1997; Weiskrantz, 1986). Blindsight has been studied in both humans and macaque monkeys and is believed to be similar in both species (Gross et al., 2004; Stoerig & Cowey, 1997; Cowey & Stoerig, 1995; Moore et al., 1995). Despite substantial efforts, the exact pathways and mechanisms responsible for blindsight are not completely resolved (Leopold, 2012).

Many studies have suggested an essential role for the SC in the mediation of blindsight (Leopold, 2012; Cowey, 2010; Yoshida et al., 2008; Weiskrantz, 2004; Gross, 1991; Vaughan & Gross, 1969). This evidence comes in large part from early lesion studies. It has long been known that ventral stream visual areas, like inferotemporal cortex, are more dependent on visual input from V1 as compared with dorsal stream areas in parietal cortex (Vaughan & Gross, 1969). Visual responses in inferotemporal cortex are abolished by V1 lesions (Rocha-Miranda, Bender, Gross, & Mishkin, 1975) and visual discriminations dependent on ventral stream processing are impaired (Cowey & Gross, 1970).

Neural responses in dorsal stream visual areas are less affected by V1 lesions. This is because dorsal stream areas are able to utilize their input from the SC in the absence of V1 (Gross, 1991). Neurons in the superior temporal polysensory area show reduced responses and selectivity after V1 lesions, but these responses break down completely after additional SC lesions (Bruce, Desimone, & Gross, 1986). Similarly, lesion or cooling of V1 alone has almost no effect on neurons in middle temporal (MT) area. However, when V1 lesions are combined with SC lesions, MT responses are eliminated (Rodman, Gross, & Albright, 1989, 1990).

How are these neurophysiological observations related to blindsight? Reversible inactivation of the SC abolishes blindsight behavior in monkeys with V1 lesions (Kato, Takaura, Ikeda, Yoshida, & Isa, 2011). Interestingly, an SC lesion alone has little or no effect on either MT or MST neuronal responses, suggesting that removal of V1 may enhance the role of the SC in visual function (Gross, 1991). Residual dorsal extrastriate activation has been observed in both monkey and human blindsight subjects using fMRI (Schmid et al., 2010; Goebel et al., 2001). Studies of more expansive cortical lesions demonstrate that activity in dorsal stream extrastriate areas is important for blindsight. Lesions that include V1 and extrastriate cortex deplete residual visual capacities and often eliminate blindsight (Leh et al., 2006; Weiskrantz, 2004; Stoerig, Faubert, Ptito, Diaconu, & Ptito, 1996). In summary, extrastriate cortex and its input from the SC are likely to play a pivotal role in blindsight (Leopold, 2012; Yoshida et al., 2008; Gross et al., 2004).

Examining the Role of SC in Blindsight Using Chromatic Stimuli

To pinpoint the role of the SC in blindsight, several investigators have attempted to block visual input to the SC by using colored stimuli. For example, the speeding of RT produced by redundant targets disappears in blindsight for both red and purple colored stimuli (Marzi et al., 2009). A similar result was found while measuring pupil and fMRI responses (Tamietto et al., 2010). The conclusion is that the SC plays an essential role in blindsight because it is selectively discriminated against by the use of colored stimuli. These data stand in contrast to several other investigations of the effects of color in blindsight.

Numerous investigators have asked whether blindsight patients can discriminate any color at all. Their results provide an apparent contradiction with studies attempting to block visual input to the SC in blindsight subjects using chromatic stimuli. Color opponency is present in human blindsight (Stoerig & Cowey, 1989, 1991), and results are similar in macaques (Cowey & Stoerig, 1999). Results in monkeys and humans reveal a role for the S-cone opponent system and suggest that, if anything, this system may be stronger in blindsight than the L-M opponent mechanism (Cowey & Stoerig, 2001). In fact, responses to green stimuli appeared to be weakest among the blue, green, and red stimuli tested. Effects of color have also been observed via fMRI activations of the SC to red, but not green, stimuli (Barbur, Sahraie, Simmons, Weiskrantz, & Williams, 1998). Together, these studies support a role for color in blindsight (Stoerig & Cowey, 1997). Nevertheless, the effects of cone contrast and luminance are difficult to resolve in these studies because stimuli were not specific to any cone or visual mechanism, specifically the S-cone mechanism.

S-cone Discrimination in Blindsight

Precise cone activation was addressed more directly using uncalibrated S-cone contrast stimuli (Leh et al., 2006). In hemispherectomized subjects, these investigators found a lack of spatial summation effect in blindsight using S-cone stimuli as compared with luminance stimuli (Leh et al., 2006). This result, combined with others using colored stimuli, suggests that color may be treated differently in blindsight than in normal vision, where V1 input to SC is available. It is difficult to say whether these results in hemispherectomized patients apply generally to blindsight because color-specific effects may depend on the presence of color selective areas in extrastriate cortex (Gross et al., 2004; Gross, 1991). Nevertheless, these results suggest that, without any cortical input available, sensitivity of SC neurons to colored stimuli is diminished.

In another effort to better isolate the S opponent mechanism, narrowband stimuli were used to address color discrimination in blindsight (Alexander & Cowey, 2010). These researchers found that color discrimination is possible, even for short wavelength stimuli. Interestingly, when the cone contrast of stimuli was specifically addressed, discrimination appeared to be better for stimuli with larger effects on the S opponent system. Contrary to evidence from hemispherectomized subjects, these data indicate that blindsight can make use of signals that preferentially excite S-cones.

Behavioral studies in monkeys provide further evidence for the utility of S-cone isolating information in blindsight. Monkeys with blindsight can make saccades to color defined targets (Yoshida et al., 2012). These targets include uncalibrated S-cone isolating stimuli, as defined in DKL color space. Although the S-cone stimulus was not individually calibrated, Yoshida et al. (2012) performed three manipulations designed to rule out stimulus detection because of artifacts. (1) Luminance contrast was varied to rule out the possibility of detection by luminance contrast. (2) Gaussian stimuli were used that eliminate the effects of presenting hard edges. (3) Luminance noise masking was used to eliminate artifacts of sudden stimulus onset, residual contrasts, and hard edge presentation. Despite efforts to ascribe blindsight detection of S-cone stimuli to other stimulus attributes, monkeys consistently demonstrated blindsight detection of S-cone stimuli. These findings are consistent with a role for chromatic detection and the SC in blindsight.

Pathways for S-cone Signals in Blindsight

What are the pathways through which S-cone signals could contribute to blindsight? From studies of neuronal degeneration patterns, it might be expected that L-M opponency would decline, whereas S opponency remained stable or strengthened (Stoerig & Cowey, 1997). After V1 lesions, there is widespread degeneration of LGN and retinal ganglion cells in pathways that carry L- and M-opponent signals, whereas those in S-cone pathways are less damaged (Cowey, Alexander, & Stoerig, 2011; Cowey, 2010). This is presumably because afferent and efferent neurons of the koniocellular layers of the LGN are preferentially spared after V1 lesions (Cowey, 2010). Not only is the koniocellular LGN less damaged by V1 lesions, neurons there may actually enhance and strengthen their role as indicated by their change in size (Leopold, 2012). Many of the neurons in LGN that survive retrograde degeneration after V1 lesions are those that project directly to extrastriate areas (Cowey & Stoerig, 1989). Direct projections from the LGN to extrastriate visual areas stem primarily from the koniocellular layers, whereas the magnocellular and parvocellular layers project predominantly to V1 (Leopold, 2012; Cowey, 2010; Hendry & Reid, 2000). The intact subcortical input from the SC also terminates predominantly in the koniocellular layers of the LGN (Leopold, 2012; Hendry & Reid, 2000).

Intriguing evidence for the role of the SC and koniocellular neurons in blindsight comes from the spatial and temporal response properties of these neurons. Koniocellular and SC neurons, which closely resemble each other (as outlined above), are similar to those most desirable for eliciting blindsight, namely, low pass filtered spatial tuning responses with sensitivity peaks around 1 cycle/degree, abrupt transient stimuli, and hard edges (Alexander & Cowey, 2010; Weiskrantz, 2004). Thus, patterns of neuronal degeneration and visual abilities after V1 lesions strongly suggest an enhanced, rather than diminished, role for S-cone pathways in blindsight.

Neuronal Pathways for Blindsight

Although the exact pathways for blindsight are not known, observations of neuronal degeneration in the LGN and retina after V1 lesions advocate a crucial role for reorganization in blindsight. Neuronal pathways for blindsight could be affected by the substantial reorganization that takes place after V1 lesions (Gross et al., 2004). Evidence for the importance of reorganization stems from V1 lesions in infancy, which generally produce more effective blindsight in monkeys (Moore, Rodman, & Gross, 2001; Moore, Rodman, Repp, Gross, & Mezrich, 1996,74). This is presumably because reorganization is faster and more effective in early life. An example of differential reorganization later in life is seen in the complex task of color discrimination. Color discrimination takes a considerable amount of time to stabilize in blindsight, suggesting that reorganization of color systems is slow to take hold (Stoerig & Cowey, 1997). Although it is difficult to reconcile the contradicting results of color discrimination and SC contributions to blindsight (described above), reorganization could be a confounding factor. Reorganization of visual systems and the koniocellular system in particular is likely to have a powerful impact on the residual visual abilities and critical structures of blindsight. Understanding how the visual system reorganizes in blindsight is critical to understanding the underlying mechanisms.

One of the most elusive aspects of the blindsight puzzle is the precise pathway(s) involved. Even after neuronal degeneration and reorganization, a number of visual pathways remain possibilities (Leopold, 2012; Stoerig & Cowey, 1997). Both the LGN and pulvinar nuclei of the thalamus project to extrastriate cortex and receive direct retinal input. Both nuclei also receive input from the SC. The SC and extrastriate visual cortex both appear to be indispensable for blindsight (Cowey, 2010; Yoshida et al., 2008; Weiskrantz, 2004; Goebel et al., 2001; Vaughan & Gross, 1969). Recent fMRI evidence in monkeys points additionally to a critical role for the LGN. Visual stimuli induce widespread activation of extrastriate cortex after V1 lesions (Schmid et al., 2010). The critical finding is that when V1 lesions are combined with inactivation of the LGN, responses in extrastriate cortex are abolished. Our data show that SC neurons are responsive to S-cone stimuli. The discovery that blindsight monkeys and blindsight patients in some studies are able to detect stimuli that preferentially activate S-cones would cast doubt on the well-established role of the SC in blindsight if the SC were not sensitive to S-cones. Our data show that, even during color discriminations, the SC could retain a central role in blindsight. Together, these findings suggest that the blindsight pathway follows a route from the SC to the koniocellular layers of the LGN to extrastriate cortex (Leopold, 2012) and can carry color signals from S-cones. This proposal completes a picture, whereby the SC, the extrastriate cortex, and the LGN are all essential to blindsight.

What Pathway(s) Carries S-cone Signals to the SC?

Our data do not address the pathway(s) that carry S-cone signals to the SC. There are two major pathways that carry visual input to the SC. Either or both of these pathways could be responsible for the observed activation of SC neurons to S-cone stimuli. It has been argued that the SC depends on the indirect pathway via LGN to V1 to generate saccades (Schiller, Kendall, Slocum, & Tehovnik, 2008). The finding that saccades can be generated to S-cone stimuli (Yoshida et al., 2012) would suggest that S-cone input can reach the SC through the indirect pathway via V1. To test whether the direct retinotectal pathway carries S-cone input to the SC would likely involve testing a function that depends on this pathway with S-cone stimuli. One example would be express saccades, which are thought to make use of the direct retinotectal pathway (Yoshida et al., 2008, but see Schiller et al., 2008, for a counter view). Under this assumption, if express saccades can be generated to S-cone isolating stimuli, then this would indicate that S-cone signals are carried in the retinotectal pathway.

Conclusions

We found that neurons in the SC respond to individually calibrated S-cone isolating stimuli on a luminance noise-masking background. The population of SC neurons is able to discriminate high from low S-cone contrasts and respond more strongly to appropriately chosen S-cone than luminance stimuli. The neuronal response latency of SC neurons correspondingly modulates with the contrast of an S-cone stimulus, with shorter latencies at higher contrasts. The sensitivity of SC neurons to changes in S-cone contrast depends on their level of transience, such that more transient neurons are more sensitive to changes in contrast. Our experimental procedures provide a direct link between our electrophysiological results and previous human psychophysical studies. Our finding that S-cone stimuli activate SC neurons rules out the possibility of blocking visual input to the SC by using S-cone isolating stimuli. Therefore, further studies are required to elucidate the neuronal foundations of and SC contributions to phenomena that have been widely studied with the use of S-cone stimuli, such as blindsight.

Acknowledgments

We would like to express our deep appreciation to Dr. Petroc Sumner for his insight and guidance with the calibration and presentation of S-cone isolating stimuli. We thank K. McCracken and Dr. Kevin Hitchens for technical assistance and our colleagues at the Center for the Neural Basis of Cognition for discussion and comments. We also thank Dr. R. Desimone for provision of the CORTEX program developed in his laboratory at NIMH. This work was supported by National Institutes of Health grants EY-12032 and MH-45156. Technical support was provided by Core grant EY-08908, and collection of MR images was supported by P41RR-03631. N. Hall was also supported by a National Science Foundation IGERT award (DGE 0549352).

Reprint requests should be sent to Nathan Hall, Department of Neuroscience, University of Pittsburgh, 115 Mellon Institute, 4400 Fifth Avenue, Pittsburgh, PA 15213, or via e-mail: njh5@cnbc.cmu.edu.

REFERENCES

REFERENCES
Alexander
,
I.
, &
Cowey
,
A.
(
2010
).
Edges, colour and awareness in blindsight.
Consciousness and Cognition
,
19
,
520
533
.
Anderson
,
E. J.
,
Husain
,
M.
, &
Sumner
,
P.
(
2008
).
Human intraparietal sulcus (IPS) and competition between exogenous and endogenous saccade plans.
Neuroimage
,
40
,
838
851
.
Anderson
,
E. J.
, &
Rees
,
G.
(
2011
).
Neural correlates of spatial orienting in the human superior colliculus.
Journal of Neurophysiology
,
106
,
2273
2284
.
Azzopardi
,
P.
, &
Cowey
,
A.
(
1997
).
Is blindsight like normal, near-threshold vision?
Proceedings of the National Academy of Sciences
,
94
,
14190
14194
.
Barbur
,
J. L.
,
Sahraie
,
A.
,
Simmons
,
A.
,
Weiskrantz
,
L.
, &
Williams
,
S. C. R.
(
1998
).
Residual processing of chromatic signals in the absence of a geniculostriate projection.
Vision Research
,
38
,
3447
3453
.
Bell
,
A. H.
,
Meredith
,
M. A.
,
Van Opstal
,
J. A.
, &
Muñoz
,
D. P.
(
2006
).
Stimulus intensity modifies saccadic reaction time and visual response latency in the superior colliculus.
Experimental Brain Research
,
174
,
53
59
.
Birch
,
J.
,
Barbur
,
J. L.
, &
Harlow
,
A. J.
(
1992
).
New method based on random luminance masking for measuring isochromatic zones using high resolution colour displays.
Ophthalmic and Physiological Optics
,
12
,
133
136
.
Boehnke
,
S. E.
,
Berg
,
D. J.
,
Marino
,
R. A.
,
Baldi
,
P. F.
,
Itti
,
L.
, &
Munoz
,
D. P.
(
2011
).
Visual adaptation and novelty responses in the superior colliculus.
European Journal of Neuroscience
,
34
,
766
779
.
Bompas
,
A.
,
Sterling
,
T.
,
Rafal
,
R. D.
, &
Sumner
,
P.
(
2008
).
Naso-temporal asymmetry for signals invisible to the retinotectal pathway.
Journal of Neurophysiology
,
100
,
412
421
.
Bompas
,
A.
, &
Sumner
,
P.
(
2008
).
Sensory sluggishness dissociates saccadic, manual, and perceptual responses: An S-cone study.
Journal of Vision
,
8
,
1
13
.
Bompas
,
A.
, &
Sumner
,
P.
(
2009
).
Oculomotor distraction by signals invisible to the retinotectal and magnocellular pathways.
Journal of Neurophysiology
,
102
,
2387
2395
.
Brainard
,
D. H.
(
1996
).
Cone contrast and opponent modulation color spaces.
In P. K. Kaiser & R. M. Boynton (Eds.)
,
Human color vision
(2nd ed., pp.
563
579
).
Washington, DC
:
Optical Society of America
.
Bruce
,
C. J.
,
Desimone
,
R.
, &
Gross
,
C. G.
(
1986
).
Both striate cortex and superior colliculus contribute to visual properties of neurons in superior temporal polysensory area of macaque monkey.
Journal of Neurophysiology
,
55
,
1057
1075
.
Bumsted
,
K.
, &
Hendrickson
,
A.
(
1999
).
Distribution and development of short-wavelength cones differ between Macaca monkey and human fovea.
The Journal of Comparative Neurology
,
403
,
502
516
.
Bunt
,
A. H.
,
Hendrickson
,
A. E.
,
Lund
,
J. S.
,
Lund
,
R. D.
, &
Fuchs
,
A. F.
(
1975
).
Monkey retinal ganglion cells: Morphometric analysis and tracing of axonal projections, with a consideration of the peroxidase technique.
The Journal of Comparative Neurology
,
164
,
265
285
.
Calkins
,
D. J.
(
2001
).
Seeing with S cones.
Progress in Retinal and Eye Research
,
20
,
255
287
.
Calkins
,
D. J.
, &
Sterling
,
P.
(
2007
).
Microcircuitry for two types of achromatic ganglion cell in primate fovea.
The Journal of Neuroscience
,
27
,
2646
2653
.
Cottaris
,
N. P.
, &
De Valois
,
R. L.
(
1998
).
Temporal dynamics of chromatic tuning in macaque primary visual cortex.
Nature
,
395
,
896
900
.
Cowey
,
A.
(
2010
).
The blindsight saga.
Experimental Brain Research
,
200
,
3
24
.
Cowey
,
A.
,
Alexander
,
I.
, &
Stoerig
,
P.
(
2011
).
Transneuronal retrograde degeneration of retinal ganglion cells and optic tract in hemianopic monkeys and humans.
Brain
,
134
,
2149
2157
.
Cowey
,
A.
, &
Gross
,
C. G.
(
1970
).
Effects of foveal prestriate and inferotemporal lesions on visual discrimination by rhesus monkeys.
Experimental Brain Research
,
11
,
128
144
.
Cowey
,
A.
, &
Stoerig
,
P.
(
1989
).
Projection patterns of surviving neurons in the dorsal lateral geniculate nucleus following discrete lesions of striate cortex: Implications for residual vision.
Experimental Brain Research
,
75
,
631
638
.
Cowey
,
A.
, &
Stoerig
,
P.
(
1995
).
Blindsight in monkeys.
Nature
,
373
,
247
249
.
Cowey
,
A.
, &
Stoerig
,
P.
(
1999
).
Spectral sensitivity in hemianopic macaque monkeys.
European Journal of Neuroscience
,
11
,
2114
2120
.
Cowey
,
A.
, &
Stoerig
,
P.
(
2001
).
Detection and discrimination of chromatic targets in hemianopic macaque monkeys and humans.
European Journal of Neuroscience
,
14
,
1320
1330
.
Crook
,
J. D.
,
Peterson
,
B. B.
,
Packer
,
O. S.
,
Robinson
,
F. R.
,
Gamlin
,
P. D.
,
Troy
,
J. B.
,
et al
(
2008
).
The smooth monostratified ganglion cell: Evidence for spatial diversity in the Y-cell pathway to the lateral geniculate nucleus and superior colliculus in the macaque monkey.
The Journal of Neuroscience
,
28
,
12654
12671
.
Crook
,
J. D.
,
Peterson
,
B. B.
,
Packer
,
O. S.
,
Robinson
,
F. R.
,
Troy
,
J. B.
, &
Dacey
,
D. M.
(
2008
).
Y-cell receptive field and collicular projection of parasol ganglion cells in macaque monkey retina.
The Journal of Neuroscience
,
28
,
11277
11291
.
Cynader
,
M.
, &
Berman
,
N.
(
1972
).
Receptive-field organization of monkey superior colliculus.
Journal of Neurophysiology
,
35
,
187
201
.
Dacey
,
D. M.
(
1996
).
Circuitry for color coding in the primate retina.
Proceedings of the National Academy of Sciences
,
93
,
582
588
.
Dacey
,
D. M.
(
1999
).
Primate retina: Cell types, circuits and color opponency.
Progress in Retinal and Eye Research
,
18
,
737
763
.
Dacey
,
D. M.
, &
Lee
,
B. B.
(
1994
).
The “blue-on” opponent pathway in primate retina originates from a distinct bistratified ganglion cell type.
Nature
,
367
,
731
735
.
Dacey
,
D. M.
,
Liao
,
H.-W.
,
Peterson
,
B. B.
,
Robinson
,
F. R.
,
Smith
,
V. C.
,
Pokorny
,
J.
,
et al
(
2005
).
Melanopsin-expressing ganglion cells in primate retina signal colour and irradiance and project to the LGN.
Nature
,
433
,
749
754
.
Dacey
,
D. M.
, &
Packer
,
O. S.
(
2003
).
Colour coding in the primate retina: Diverse cell types and cone-specific circuitry.
Current Opinion in Neurobiology
,
13
,
421
427
.
Dacey
,
D. M.
,
Peterson
,
B. B.
,
Robinson
,
F. R.
, &
Gamlin
,
P. D.
(
2003
).
Fireworks in the primate retina: In vitro photodynamics reveals diverse LGN-projecting ganglion cell types.
Neuron
,
37
,
15
27
.
De Monasterio
,
F. M.
(
1978a
).
Properties of concentrically organized X and Y ganglion cells of macaque retina.
Journal of Neurophysiology
,
41
,
1394
1417
.
De Monasterio
,
F. M.
(
1978b
).
Properties of ganglion cells with atypical receptive-field organization in retina of macaques.
Journal of Neurophysiology
,
41
,
1435
1449
.
De Monasterio
,
F. M.
, &
Gouras
,
P.
(
1975
).
Functional properties of ganglion cells of the rhesus monkey retina.
The Journal of Physiology
,
251
,
167
195
.
Derrington
,
A. M.
(
2002
).
Visual system: S is not for saccades.
Current Biology: CB
,
12
,
R591
R592
.
Derrington
,
A. M.
,
Krauskopf
,
J.
, &
Lennie
,
P.
(
1984
).
Chromatic mechanisms in lateral geniculate nucleus of macaque.
The Journal of Physiology
,
357
,
241
265
.
DuBois
,
R. M.
, &
Cohen
,
M. S.
(
2000
).
Spatiotopic organization in human superior colliculus observed with fMRI.
Neuroimage
,
12
,
63
70
.
Dunn
,
C. A.
,
Hall
,
N. J.
, &
Colby
,
C. L.
(
2010
).
Spatial updating in monkey superior colliculus in the absence of the forebrain commissures: Dissociation between superficial and intermediate layers.
Journal of Neurophysiology
,
104
,
1267
1285
.
Finlay
,
B. L.
,
Schiller
,
P. H.
, &
Volman
,
S. F.
(
1976
).
Quantitative studies of single-cell properties in monkey striate cortex. IV. Corticotectal cells.
Journal of Neurophysiology
,
39
,
1352
1361
.
Fries
,
W.
(
1984
).
Cortical projections to the superior colliculus in the macaque monkey: A retrograde study using horseradish peroxidase.
Journal of Comparative Neurology
,
230
,
55
76
.
Goebel
,
R.
,
Muckli
,
L.
,
Zanella
,
F. E.
,
Singer
,
W.
, &
Stoerig
,
P.
(
2001
).
Sustained extrastriate cortical activation without visual awareness revealed by fMRI studies of hemianopic patients.
Vision Research
,
41
,
1459
1474
.
Goldberg
,
M. E.
, &
Wurtz
,
R. H.
(
1972
).
Activity of superior colliculus in behaving monkey. I. Visual receptive fields of single neurons.
Journal of Neurophysiology
,
35
,
542
559
.
Gross
,
C. G.
(
1991
).
Contribution of striate cortex and the superior colliculus to visual function in area MT, the superior temporal polysensory area and inferior temporal cortex.
Neuropsychologia
,
29
,
497
515
.
Gross
,
C. G.
,
Moore
,
T.
, &
Rodman
,
H. R.
(
2004
).
Visually guided behavior after V1 lesions in young and adult monkeys and its relation to blindsight in humans.
Progress in Brain Research
,
144
,
279
294
.
Hall
,
N. J.
, &
Colby
,
C. L.
(
2013
).
Psychophysical definition of S-cone stimuli in the macaque.
Journal of Vision
,
13
,
article 20
.
Hendrickson
,
A.
,
Wilson
,
M. E.
, &
Toyne
,
M. J.
(
1970
).
The distribution of optic nerve fibers in Macaca mulatta.
Brain Research
,
23
,
425
427
.
Hendry
,
S. H. C.
, &
Reid
,
R. C.
(
2000
).
The koniocellular pathway in primate vision.
Annual Review of Neuroscience
,
23
,
127
153
.
Hovland
,
C.
, &
Bradshaw
,
D.
(
1937
).
Visual reaction time as a function of stimulus background contrast.
Psychologische Forschung
,
21
,
50
55
.
Hubel
,
D. H.
,
LeVay
,
S.
, &
Wiesel
,
T. N.
(
1975
).
Mode of termination of retinotectal fibers in macaque monkey: An autoradiographic study.
Brain Research
,
96
,
25
40
.
Jacobs
,
G. H.
(
1998
).
A perspective on color vision in platyrrhine monkeys.
Vision Research
,
38
,
3307
3313
.
Jacobs
,
G.
(
2007
).
New World monkeys and color.
International Journal of Primatology
,
28
,
729
759
.
Joo
,
H. R.
,
Peterson
,
B. B.
,
Haun
,
T. J.
, &
Dacey
,
D. M.
(
2011
).
Characterization of a novel large-field cone bipolar cell type in the primate retina: Evidence for selective cone connections.
Visual Neuroscience
,
28
,
29
37
.
Judge
,
S.
,
Richmond
,
B.
, &
Chu
,
F.
(
1980
).
Implantation of magnetic search coils for measurement of eye position: An improved method.
Vision Research
,
20
,
535
538
.
Kaas
,
J. H.
(
2004
).
The evolution of the visual system in primates.
In L. M. Chalupa & J. S. Werner (Eds.)
,
The visual neurosciences
(
Vol. 1
, pp.
1563
1572
).
Cambridge, MA
:
MIT Press
.
Kato
,
R.
,
Takaura
,
K.
,
Ikeda
,
T.
,
Yoshida
,
M.
, &
Isa
,
T.
(
2011
).
Contribution of the retino-tectal pathway to visually guided saccades after lesion of the primary visual cortex in monkeys.
European Journal of Neuroscience
,
33
,
1952
1960
.
Klug
,
K.
,
Herr
,
S.
,
Ngo
,
I. T.
,
Sterling
,
P.
, &
Schein
,
S.
(
2003
).
Macaque retina contains an S-cone OFF midget pathway.
The Journal of Neuroscience
,
23
,
9881
9887
.
Kuypers
,
H. G. J. M.
, &
Lawrence
,
D. G.
(
1967
).
Cortical projections to the red nucleus and the brain stem in the rhesus monkey.
Brain Research
,
4
,
151
188
.
Lee
,
J.
,
Williford
,
T.
, &
Maunsell
,
J. H. R.
(
2007
).
Spatial attention and the latency of neuronal responses in macaque area V4.
The Journal of Neuroscience
,
27
,
9632
9637
.
Leh
,
S. E.
,
Mullen
,
K. T.
, &
Ptito
,
A.
(
2006
).
Absence of S-cone input in human blindsight following hemispherectomy.
European Journal of Neuroscience
,
24
,
2954
2960
.
Leh
,
S. E.
,
Ptito
,
A.
,
Schönwiesner
,
M.
,
Chakravarty
,
M. M.
, &
Mullen
,
K. T.
(
2010
).
Blindsight mediated by an S-cone-independent collicular pathway: An fMRI study in hemispherectomized subjects.
Journal of Cognitive Neuroscience
,
22
,
670
682
.
Leo
,
F.
,
Bertini
,
C.
,
di Pellegrino
,
G.
, &
La davas
,
E.
(
2008
).
Multisensory integration for orienting responses in humans requires the activation of the superior colliculus.
Experimental Brain Research
,
186
,
67
77
.
Leopold
,
D. A.
(
2012
).
Primary visual cortex: Awareness and blindsight*.
Annual Review of Neuroscience
,
35
,
91
109
.
Levenson
,
D. H.
,
Fernandez-duque
,
E.
,
Evans
,
S.
, &
Jacobs
,
G. H.
(
2007
).
Mutational changes in S-cone opsin genes common to both nocturnal and cathemeral Aotus monkeys.
American Journal of Primatology
,
69
,
757
765
.
Leventhal
,
A. G.
,
Rodieck
,
R. W.
, &
Dreher
,
B.
(
1981
).
Retinal ganglion cell classes in the Old World monkey: Morphology and central projections.
Science
,
213
,
1139
1142
.
MacLeod
,
D. I. A.
, &
Boynton
,
R. M.
(
1979
).
Chromaticity diagram showing cone excitation by stimuli of equal luminance.
Journal of the Optical Society of America
,
69
,
1183
1186
.
Marrocco
,
R. T.
, &
Li
,
R. H.
(
1977
).
Monkey superior colliculus: Properties of single cells and their afferent inputs.
Journal of Neurophysiology
,
40
,
844
860
.
Martin
,
P. R.
,
White
,
A. J. R.
,
Goodchild
,
A. K.
,
Wilder
,
H. D.
, &
Sefton
,
A. E.
(
1997
).
Evidence that blue-on cells are part of the third geniculocortical pathway in primates.
European Journal of Neuroscience
,
9
,
1536
1541
.
Marzi
,
C. A.
,
Mancini
,
F.
,
Metitieri
,
T.
, &
Savazzi
,
S.
(
2009
).
Blindsight following visual cortex deafferentation disappears with purple and red stimuli: A case study.
Neuropsychologia
,
47
,
1382
1385
.
Maunsell
,
J. H.
, &
Gibson
,
J. R.
(
1992
).
Visual response latencies in striate cortex of the macaque monkey.
Journal of Neurophysiology
,
68
,
1332
1344
.
Mollon
,
J. D.
(
1989
).
Tho' she kneel'd in that place where they grew: The uses and origins of primate colour vision.
Journal of Experimental Biology
,
146
,
21
38
.
Mollon
,
J. D.
, &
Polden
,
P. G.
(
1975
).
Colour illusion and evidence for interaction between cone mechanisms.
Nature
,
258
,
421
422
.
Moore
,
T.
,
Rodman
,
H. R.
, &
Gross
,
C. G.
(
2001
).
Direction of motion discrimination after early lesions of striate cortex (V1) of the macaque monkey.
Proceedings of the National Academy of Sciences, U.S.A.
,
98
,
325
330
.
Moore
,
T.
,
Rodman
,
H. R.
,
Repp
,
A. B.
, &
Gross
,
C. G.
(
1995
).
Localization of visual stimuli after striate cortex damage in monkeys: Parallels with human blindsight.
Proceedings of the National Academy of Sciences
,
92
,
8215
8218
.
Moore
,
T.
,
Rodman
,
H. R.
,
Repp
,
A. B.
,
Gross
,
C. G.
, &
Mezrich
,
R. S.
(
1996
).
Greater residual vision in monkeys after striate cortex damage in infancy.
Journal of Neurophysiology
,
76
,
3928
3933
.
Nassi
,
J. J.
, &
Callaway
,
E. M.
(
2009
).
Parallel processing strategies of the primate visual system.
Nature Reviews Neuroscience
,
10
,
360
372
.
Nathans
,
J.
,
Thomas
,
D.
, &
Hogness
,
D. S.
(
1986
).
Molecular genetics of human color vision: The genes encoding blue, green, and red pigments.
Science
,
232
,
193
202
.
Perry
,
V. H.
, &
Cowey
,
A.
(
1984
).
Retinal ganglion cells that project to the superior colliculus and pretectum in the macaque monkey.
Neuroscience
,
12
,
1125
1137
.
Rocha-Miranda
,
C. E.
,
Bender
,
D. B.
,
Gross
,
C. G.
, &
Mishkin
,
M.
(
1975
).
Visual activation of neurons in inferotemporal cortex depends on striate cortex and forebrain commissures.
Journal of Neurophysiology
,
38
,
475
491
.
Rodieck
,
R. W.
, &
Watanabe
,
M.
(
1993
).
Survey of the morphology of macaque retinal ganglion cells that project to the pretectum, superior colliculus, and parvicellular laminae of the lateral geniculate nucleus.
The Journal of Comparative Neurology
,
338
,
289
303
.
Rodman
,
H. R.
,
Gross
,
C. G.
, &
Albright
,
T. D.
(
1989
).
Afferent basis of visual response properties in area MT of the macaque. I. Effects of striate cortex removal.
The Journal of Neuroscience
,
9
,
2033
2050
.
Rodman
,
H. R.
,
Gross
,
C. G.
, &
Albright
,
T. D.
(
1990
).
Afferent basis of visual response properties in area MT of the macaque. II. Effects of superior colliculus removal.
The Journal of Neuroscience
,
10
,
1154
1164
.
Sanders
,
M. D.
,
Warrington
,
E.
,
Marshall
,
J.
, &
Wieskrantz
,
L.
(
1974
).
“Blindsight“: Vision in a field defect.
The Lancet
,
303
,
707
708
.
Savazzi
,
S.
,
Fabri
,
M.
,
Rubboli
,
G.
,
Paggi
,
A.
,
Tassinari
,
C. A.
, &
Marzi
,
C. A.
(
2007
).
Interhemispheric transfer following callosotomy in humans: Role of the superior colliculus.
Neuropsychologia
,
45
,
2417
2427
.
Savazzi
,
S.
, &
Marzi
,
C. A.
(
2004
).
The superior colliculus subserves interhemispheric neural summation in both normals and patients with a total section or agenesis of the corpus callosum.
Neuropsychologia
,
42
,
1608
1618
.
Schiller
,
P. H.
, &
Colby
,
C. L.
(
1983
).
The responses of single cells in the lateral geniculate nucleus of the rhesus monkey to color and luminance contrast.
Vision Research
,
23
,
1631
1641
.
Schiller
,
P. H.
,
Kendall
,
G. L.
,
Slocum
,
W. M.
, &
Tehovnik
,
E. J.
(
2008
).
Conditions that alter saccadic eye movement latencies and affect target choice to visual stimuli and to electrical stimulation of area V1 in the monkey.
Visual Neuroscience
,
25
,
661
673
.
Schiller
,
P. H.
, &
Koerner
,
F.
(
1971
).
Discharge characteristics of single units in superior colliculus of the alert rhesus monkey.
Journal of Neurophysiology
,
34
,
920
936
.
Schiller
,
P. H.
, &
Malpeli
,
J. G.
(
1977
).
Properties and tectal projections of monkey retinal ganglion cells.
Journal of Neurophysiology
,
40
,
428
445
.
Schiller
,
P. H.
,
Malpeli
,
J. G.
, &
Schein
,
S. J.
(
1979
).
Composition of geniculostriate input to superior colliculus of the rhesus monkey.
Journal of Neurophysiology
,
42
,
1124
1133
.
Schiller
,
P. H.
, &
Stryker
,
M.
(
1972
).
Single-unit recording and stimulation in superior colliculus of the alert rhesus monkey.
Journal of Neurophysiology
,
35
,
915
924
.
Schiller
,
P. H.
,
Stryker
,
M.
,
Cynader
,
M.
, &
Berman
,
N.
(
1974
).
Response characteristics of single cells in the monkey superior colliculus following ablation or cooling of visual cortex.
Journal of Neurophysiology
,
37
,
181
194
.
Schmid
,
M. C.
,
Mrowka
,
S. W.
,
Turchi
,
J.
,
Saunders
,
R. C.
,
Wilke
,
M.
,
Peters
,
A. J.
,
et al
(
2010
).
Blindsight depends on the lateral geniculate nucleus.
Nature
,
466
,
373
377
.
Schmolesky
,
M. T.
,
Wang
,
Y.
,
Hanes
,
D. P.
,
Thompson
,
K. G.
,
Leutgeb
,
S.
,
Schall
,
J. D.
,
et al
(
1998
).
Signal timing across the macaque visual system.
Journal of Neurophysiology
,
79
,
3272
3278
.
Schneider
,
K. A.
, &
Kastner
,
S.
(
2005
).
Visual responses of the human superior colliculus: A high-resolution functional magnetic resonance imaging study.
Journal of Neurophysiology
,
94
,
2491
2503
.
Silveira
,
L. C. L.
,
Lee
,
B. B.
,
Yamada
,
E. S.
,
Kremers
,
J. A. N.
,
Hunt
,
D. M.
,
Martin
,
P. R.
,
et al
(
1999
).
Ganglion cells of a short-wavelength-sensitive cone pathway in New World monkeys: Morphology and physiology.
Visual Neuroscience
,
16
,
333
343
.
Smith
,
V. C.
, &
Pokorny
,
J.
(
1975
).
Spectral sensitivity of the foveal cone photopigments between 400 and 500 nm.
Vision Research
,
15
,
161
171
.
Smithson
,
H. E.
, &
Mollon
,
J. D.
(
2004
).
Is the S-opponent chromatic sub-system sluggish?
Vision Research
,
44
,
2919
2929
.
Smithson
,
H. E.
,
Sumner
,
P.
, &
Mollon
,
J. D.
(
2003
).
How to find a tritan line.
In J. D. Mollon, J. Pokorny, & K. Knoblauch (Eds.)
,
Normal and defective colour vision
(pp.
279
287
).
Oxford, UK
:
Oxford University Press
.
Snodderly
,
D. M.
,
Auran
,
J. D.
, &
Delori
,
F. C.
(
1984
).
The macular pigment. II. Spatial distribution in primate retinas.
Investigative Ophthalmology & Visual Science
,
25
,
674
685
.
Snodderly
,
D. M.
,
Brown
,
P. K.
,
Delori
,
F. C.
, &
Auran
,
J. D.
(
1984
).
The macular pigment. I. Absorbance spectra, localization, and discrimination from other yellow pigments in primate retinas.
Investigative Ophthalmology & Visual Science
,
25
,
660
673
.
Sommer
,
M. A.
, &
Wurtz
,
R. H.
(
2004
).
The dialogue between cerebral cortex and superior colliculus: Implications for saccadic target selection and corollary discharge.
In L. M. Chalupa & J. S. Werner (Eds.)
,
The visual neurosciences
(
Vol. 2
, pp.
1466
1484
).
Cambridge, MA
:
MIT Press
.
Stoerig
,
P.
, &
Cowey
,
A.
(
1989
).
Wavelength sensitivity in blindsight.
Nature
,
342
,
916
918
.
Stoerig
,
P.
, &
Cowey
,
A.
(
1991
).
Increment-threshold spectral sensitivity in blindsight: Evidence for colour opponency.
Brain
,
114
,
1487
1512
.
Stoerig
,
P.
, &
Cowey
,
A.
(
1997
).
Blindsight in man and monkey.
Brain
,
120
,
535
559
.
Stoerig
,
P.
,
Faubert
,
J.
,
Ptito
,
M.
,
Diaconu
,
V.
, &
Ptito
,
A.
(
1996
).
No blindsight following hemidecortication in human subjects?
NeuroReport
,
7
,
1990
1994
.
Sumner
,
P.
(
2006
).
Inhibition versus attentional momentum in cortical and collicular mechanisms of IOR.
Cognitive Neuropsychology
,
23
,
1035
1048
.
Sumner
,
P.
,
Adamjee
,
T.
, &
Mollon
,
J. D.
(
2002
).
Signals invisible to the collicular and magnocellular pathways can capture visual attention.
Current Biology
,
12
,
1312
1316
.
Sumner
,
P.
,
Nachev
,
P.
,
Castor-Perry
,
S.
,
Isenman
,
H.
, &
Kennard
,
C.
(
2006
).
Which visual pathways cause fixation-related inhibition?
Journal of Neurophysiology
,
95
,
1527
1536
.
Sumner
,
P.
,
Nachev
,
P.
,
Vora
,
N.
,
Husain
,
M.
, &
Kennard
,
C.
(
2004
).
Distinct cortical and collicular mechanisms of inhibition of return revealed with S cone stimuli.
Current Biology
,
14
,
2259
2263
.
Szmajda
,
B. A.
,
Buzas
,
P.
,
FitzGibbon
,
T.
, &
Martin
,
P. R.
(
2006
).
Geniculocortical relay of blue-off signals in the primate visual system.
Proceedings of the National Academy of Sciences
,
103
,
19512
19517
.
Szmajda
,
B. A.
,
Grünert
,
U.
, &
Martin
,
P. R.
(
2008
).
Retinal ganglion cell inputs to the koniocellular pathway.
The Journal of Comparative Neurology
,
510
,
251
268
.
Tailby
,
C.
,
Cheong
,
S. K.
,
Pietersen
,
A. N.
,
Solomon
,
S. G.
, &
Martin
,
P. R.
(
2012
).
Colour and pattern selectivity of receptive fields in superior colliculus of marmoset monkeys.
The Journal of Physiology
,
590
,
4061
4077
.
Tailby
,
C.
,
Szmajda
,
B. A.
,
Buzas
,
P.
,
Lee
,
B. B.
, &
Martin
,
P. R.
(
2008
).
Transmission of blue (S) cone signals through the primate lateral geniculate nucleus.
The Journal of Physiology
,
586
,
5947
5967
.
Tamietto
,
M.
,
Cauda
,
F.
,
Corazzini
,
L. L.
,
Savazzi
,
S.
,
Marzi
,
C. A.
,
Goebel
,
R.
,
et al
(
2010
).
Collicular vision guides nonconscious behavior.
Journal of Cognitive Neuroscience
,
22
,
888
902
.
Thirkettle
,
M.
,
Walton
,
T.
,
Shah
,
A.
,
Gurney
,
K.
,
Redgrave
,
P.
, &
Stafford
,
T.
(
2013
).
The path to learning: Action acquisition is impaired when visual reinforcement signals must first access cortex.
Behavioural Brain Research
,
243
,
267
272
.
Vaughan
,
H. G.
, Jr., &
Gross
,
C. G.
(
1969
).
Cortical responses to light in unanesthetized monkeys and their alteration by visual system lesions.
Experimental Brain Research
,
8
,
19
36
.
Weiskrantz
,
L.
(
1986
).
Blindsight. A case study and implications.
Oxford, UK
:
Oxford University Press
.
Weiskrantz
,
L.
(
2004
).
Blindsight.
In L. M. Chalupa & J. S. Werner (Eds.)
,
The visual neurosciences
(
Vol. 1
, pp.
657
669
).
Cambridge, MA
:
MIT Press
.
White
,
B. J.
,
Boehnke
,
S. E.
,
Marino
,
R. A.
,
Itti
,
L.
, &
Munoz
,
D. P.
(
2009
).
Color-related signals in the primate superior colliculus.
The Journal of Neuroscience
,
29
,
12159
12166
.
Yokoyama
,
S.
, &
Yokoyama
,
R.
(
1989
).
Molecular evolution of human visual pigment genes.
Molecular Biology and Evolution
,
6
,
186
197
.
Yoshida
,
M.
,
Itti
,
L.
,
Berg
,
D. J.
,
Ikeda
,
T.
,
Kato
,
R.
,
Takaura
,
K.
,
et al
(
2012
).
Residual attention guidance in blindsight monkeys watching complex natural scenes.
Current Biology
,
22
,
1429
1434
.
Yoshida
,
M.
,
Takaura
,
K.
,
Kato
,
R.
,
Ikeda
,
T.
, &
Isa
,
T.
(
2008
).
Striate cortical lesions affect deliberate decision and control of saccade: Implication for blindsight.
The Journal of Neuroscience
,
28
,
10517
10530
.