Abstract

Perception of known patterns results from the interaction of current sensory input with existing internal representations. It is unclear how perceptual and mnemonic processes interact when visual input is dynamic and structured such that it does not allow immediate recognition of obvious objects and forms. In an fMRI experiment, meaningful visual motion stimuli depicting movement through a virtual tunnel and indistinct, meaningless visual motion stimuli, achieved through phase scrambling of the same stimuli, were presented while participants performed an optic flow task. We found that our indistinct visual motion stimuli evoked hippocampal activation, whereas the corresponding meaningful stimuli did not. Using independent component analysis, we were able to demonstrate a functional connectivity between the hippocampus and early visual areas, with increased activity for indistinct stimuli. In a second experiment, we used the same stimuli to test whether our results depended on the participants' task. We found task-independent bilateral hippocampal activation in response to indistinct motion stimuli. For both experiments, psychophysiological interaction analysis revealed a coupling from posterior hippocampus to dorsal visuospatial and ventral visual object processing areas when viewing indistinct stimuli. These results indicate a close functional link between stimulus-dependent perceptual and mnemonic processes. The observed pattern of hippocampal functional connectivity, in the absence of an explicit memory task, suggests that cortical–hippocampal networks are recruited when visual stimuli are temporally uncertain and do not immediately reveal a clear meaning.

INTRODUCTION

When exploring our environment, we are confronted with dynamic visual input that is recognized by reference to existing memories and concepts. Although recognition and categorization of images occurs quickly and without conscious effort, it is an open question how indistinct dynamic visual stimuli are processed, for which a meaning is not immediately apparent and which cannot be easily categorized. One brain area of major importance for the creation, retrieval, and manipulation of explicit memories and concepts is the hippocampus and the medial-temporal lobe (Carr, Rissman, & Wagner, 2010; Eldridge, Engel, Zeineh, Bookheimer, & Knowlton, 2005; Eichenbaum, 2004). The hippocampus represents a convergence zone for multiple sensory inputs, receiving highly integrated information from the association cortices of the respective sensory regions (Buckner, 2010; Burwell & Agster, 2008; Lavenex & Amaral, 2000; Amaral & Witter, 1989), and would thus be a candidate area involved in processing indistinct visual stimuli.

For static images, several studies have addressed which brain regions are active for meaningful and recognizable stimuli versus meaningless stimuli. One report found that the hippocampus was active as participants viewed meaningful scenes when compared with meaningless, scrambled scenes (Binder, Bellgowan, Hammeke, Possing, & Frost, 2005). Visual discriminations between meaningful (known) everyday objects relative to unknown novel objects have also been found to activate the posterior hippocampus bilaterally (Barense, Henson, & Graham, 2011). Furthermore, visual noise stimuli do not evoke hippocampal activation (Martin, 1999), and studies investigating static phase-manipulated images have not reported hippocampal activation (Wichmann, Braun, & Gegenfurtner, 2006; Olman, Ugurbil, Schrater, & Kersten, 2004). Thus, based on these studies using static visual input, one would not expect to find hippocampal activation in response to meaningless visual input.

Despite the evidence accumulated with static images, the cognitive processing demands may differ significantly for dynamic visual input that neither immediately reveals a clear meaning nor can be categorized as noise. Object and scene recognition occurs naturally in a dynamic environment, and it appears that the temporal dimension is critical for our ability to recognize objects independent of size, location, and viewing angle (Li & DiCarlo, 2010). In a first step toward understanding the mechanisms underlying the processing of this type of dynamic visual input, we wanted to clarify which brain areas are involved in such situations and investigate their functional connectivity.

A recent perspective on human perception suggests that visual recognition requires linking current visual input to a corresponding memory representation (Bar, 2009). Retrieval of incomplete or degraded sensory cues can be accomplished through pattern completion, a mechanism by which a stored memory trace can be retrieved through hippocampal recurrent connections (Bird & Burgess, 2008; Norman & O'Reilly, 2003; Rudy & O'Reilly, 2001, Levy, 1996). Thus, perception of visual input that is difficult to recognize may elicit retrieval processes of stored memory representations.

The hippocampus has also been associated with imagination (Buckner, 2010) and implicated in a network for making predictions (Bar, 2009; Schacter & Addis, 2009). Computational models have suggested that predictions are automatically compared with sensory input to detect if the environmental input represents a mismatch to the expectation (Lisman & Grace, 2005; Hasselmo, Schnell, & Barkai, 1995).

By definition, indistinct visual motion input is structured such that no obvious objects and forms are contained and that constantly changes its appearance; therefore, correct predictions should be harder to make for indistinct visual motion stimuli than for clearly structured visual motion. The comparison between indistinct visual input and the expectation should evoke a continuous mismatch. In human imaging studies, mismatch has shown hippocampal activity that scales with the number of changes in the environment (Duncan, Ketz, Inati, & Davachi, 2012). The mismatch signal may be essential for encoding to ensure the accuracy of subsequent predictions. Similarly, one study that investigated a context-specific form of novelty processing found entropy or expected uncertainty of events, in particular, contextual uncertainty for visual stimuli, to be associated with hippocampal activation (Strange, Duggins, Penny, Dolan, & Friston, 2005).

We thus investigated brain activity in response to meaningful moving stimuli, which can be easily categorized, compared with indistinct moving stimuli, for which a category is hard to find. Meaningful stimuli were emotionally neutral virtual tunnels that represent self-motion in space, whereas indistinct visual input was constructed by phase scrambling the meaningful visual stimuli as it renders the stimuli that are hard to categorize although distinct from visual noise. Because the alignment of phase information is essential for recognizing edges and spatial structure in images (e.g., Wichmann et al., 2006), phase scrambling of the tunnel films created new stimuli, which are comparable in terms of image statistics but do not contain recognizable features such as edges or structural information. Previously, we had shown that these stimuli evoked strong activation in early visual areas (Fraedrich, Glasauer, & Flanagin, 2010). For the current study, we investigated in a first step the functional connectivity of the previously found visual areas using independent component analysis (ICA) to determine how the network responds when participants cannot predict upcoming visual scenes. In an additional task paradigm, we tested whether stimulus- or task-dependent hippocampal activation exists. This study seeks to clarify whether the hippocampus is recruited in response to dynamic indistinct stimuli. If the hippocampus is involved in processing such stimuli, then not only the activity of the hippocampus but also its functional coupling is expected to be stimulus-dependent. This was tested in a final step using psychophysiological interaction (PPI) analysis (Friston et al., 1997).

METHODS

Experiment 1

Participants

Twenty-nine right-handed healthy young volunteers (17 men, mean age = 25.0 years, SD = 2.05 years) with normal or corrected-to-normal vision and no documented history of neurological or psychiatric history gave their informed consent to participate in the study. The experiment was conducted in accordance with the Declaration of Helsinki and approved by the local ethics committee of the medical faculty at the Ludwig–Maximilians University, Munich.

Stimuli and Experimental Procedure

The original visual motion stimuli comprised 12 different 6-sec virtual tunnels (created with Open GL) consisting of a straight, a curved, and another straight segment (resolution = 550 × 549) with varying turn angles of 30°, 40°, 70°, 80°, 110°, and 120° to the left or right. These stimuli were Fourier-transformed using the discrete 3-D Fourier transform implemented in Matlab (MathWorks, Natick, MA). The phase components of all frequencies were then randomly exchanged, and the signal was back-transformed into the time domain, resulting in the phase-scrambled stimuli. Corresponding stimuli had the same amount of local flow, average contrast, and luminance, but phase-scrambled stimuli had larger frame-wise local luminance changes. As such phase scrambling maintained the spatial and temporal visual motion statistics relevant for low-level visual processing and eliminated the presence of any obvious form, edge, or structure. We therefore refer to phase-scrambled stimuli as indistinct stimuli and tunnel stimuli as meaningful. Stimulus presentation was followed by a 3-sec response interval in which participants were instructed to indicate, with a button press, the main direction of optic flow motion (left or right, Figure 1A). The duration between onset of response interval and button press is referred to as response time. The optic flow motion task will from now on be termed as direction task. Stimuli were presented in pairs of the same stimulus type such that one stimulus block was 18 sec, and the blocks alternated between stimulus type. Forty-eight stimuli (24 for each film type) were presented in each run, for a total of 12 blocks per stimulus per run (192 scans). Two runs were acquired per participant.

Figure 1. 

Experimental design and stimuli. (A) The alternating sequence of two indistinct (phase-scrambled) and two meaningful (tunnel) stimuli, repeated over the length of the experiment. Each film was presented for 6 sec and followed by a 3-sec response period, here exemplified with the response choice of the direction task, translated from German. In Experiment 2, each task was presented in a single block preceded by a six-scan fixation cross and a three-scan instruction screen to prevent task switching effects. (B, C) Example frames from the detection task containing the to-be-detected arrows for phase-scrambled stimuli with the arrow in the top left quadrant (B) and for tunnel stimuli with an arrow in the bottom left quadrant (C). The arrows in the insets have been enhanced in their red color to make the arrow more visible to the reader (original saturation in whole frame). Also, note the red fixation cross in both frames.

Figure 1. 

Experimental design and stimuli. (A) The alternating sequence of two indistinct (phase-scrambled) and two meaningful (tunnel) stimuli, repeated over the length of the experiment. Each film was presented for 6 sec and followed by a 3-sec response period, here exemplified with the response choice of the direction task, translated from German. In Experiment 2, each task was presented in a single block preceded by a six-scan fixation cross and a three-scan instruction screen to prevent task switching effects. (B, C) Example frames from the detection task containing the to-be-detected arrows for phase-scrambled stimuli with the arrow in the top left quadrant (B) and for tunnel stimuli with an arrow in the bottom left quadrant (C). The arrows in the insets have been enhanced in their red color to make the arrow more visible to the reader (original saturation in whole frame). Also, note the red fixation cross in both frames.

The data collected in the first experiment were also used to examine the visual response to phase manipulation in the spatial and temporal domain and have been published as such elsewhere (Fraedrich et al., 2010). These data were reanalyzed here for functional connectivity and activity in the hippocampus. We will describe these additional analyses here, only summarizing the relevant methodological details. The experiment is described in detail in Fraedrich et al. (2010). The MR data acquisition parameters and visual projection of the stimuli are the same as in Experiment 2 and can be found there.

Mutual Information for Both Stimulus Types

For each stimulus type (meaningful and indistinct stimuli), the mutual information was computed to assess the degree to which one frame contains information about the next one. The mutual information Ii between the current frame Xi and the previous frame Xi−1 was computed from the entropy H(X) of a frame according to the formula
formula
with H(Xi,Xi−1) being the joint entropy of the two adjacent frames. Computations were performed using Matlab (MathWorks). This was done for all frames of each film, and the resulting mutual information was then averaged separately over allmeaningful (n = 12) and indistinct films (n = 12). The average mutual information was significantly higher (t test, p < .0001) for all meaningful stimuli (mean = 3.25 bits, SD = 0.05 bits) than for indistinct stimuli (mean= 2.27 bits, SD = 0.18 bits).

Functional Connectivity Using ICA

ICA models functional MRI data as linear mixtures of spatially independent processes, each contributing to the data set with an unknown time profile and as such provides a measure of functional connectivity between discrete brain regions (Greicius & Menon, 2004; Van de Ven, Formisano, Prvulovic, Roeder, & Linden, 2004). We performed ICA on the 18 participants from the main experiment in Fraedrich et al. (2010) using FMRLAB 4.0 (Duann et al., 2002) for Matlab (MathWorks). We used only the participants who did not see the fixation cross because we wanted to look at the network connectivity under the most natural viewing conditions. Each image was slice-time corrected, minimizing the differences in light intensity because of acquisition timing. Non-brain image voxels were removed from further analysis by masking the fMRI time series images with the intensity-thresholded structural images. The image time series was quadratically normalized, the temporal and voxel means were removed, and then the runs were concatenated for the analysis. The Infomax algorithm (McKeown et al., 1998; Bell & Sejnowski, 1995) was used to separate components by maximizing the kurtosis of the components. The ICA unmixing matrix was computed using the runica routine (Matlab version; Makeig, Bell, Jung, & Sejnowski, 1996). After convergence, 160 spatially independent component maps were derived for each participant, normalized by subtracting the component map's mean from each voxel and dividing by the standard deviation of the map weights. Because of the mostly super-Gaussian nature of independent components, each component map or region of activity (ROA) comprised all voxels with z values above 1.5. Structural images and component maps were normalized to the standard Talairach space using SPM2 (Wellcome Trust, London, United Kingdom).

To select equivalent independent components across participants, two independent observers labeled the artifact-free ICs based on visual inspection of their spatial ROAs, their consistency of brain activation, and the computed overlap ratio between components (overlap ratio = same voxels highlighted as ROA voxels across participants). ROA of each selected component was visualized within a high-resolution structural image (MRIcron anatomic template ch2bet) for comparison across participants. The independent component for early visual areas from each participant was solely selected based on their ROAs in early visual areas. We tested for participation of all brain areas in the visual component on the group level by applying a one-sample t test over the normalized ROAs, thresholding at p < .05, false discovery rate (FDR) corrected for multiple comparisons. To investigate whether this component is related to stimulus presentation, the back-projected event-related BOLD responses were computed from the visual independent component for all participants. First, a more specific time course of each component was computed from the highly participating voxels (z threshold of 4.0) in the component ROA. The component time course was then epoched from 4.5 sec before to 18 sec after the stimulus presentation, resulting in 48 22.5-sec epochs. The event-related BOLD response and its standard deviation were computed across epochs of the same event types. Because no explicit baseline measure was acquired, the component time course was compared against a baseline measure obtained from using a bootstrapping approach. This baseline was computed from 100 epochs of equal length that were randomly picked from the entire back-projected component time course and averaged.

Univariate Analysis

All participants were reanalyzed for hippocampal activity using a hierarchical general linear model. fMRI preprocessing and statistical analyses were conducted using SPM5 (Wellcome Trust, London, United Kingdom). EPI data were realigned using a six-parameter rigid body transformation, spatially normalized to Montreal Neurological Institute space, and smoothed with an isotropic 8-mm FWHM Gaussian filter. To remove low-frequency noise and slow drifts in the signal, a high-pass filter (cutoff = 128 sec corresponding to 0.0078 Hz) was included in the filtering matrix. On the single-subject level, regressors for both types of stimuli and the participant responses as well as six motion correction parameters (as effects of no interest) were used to model the data. Contrasts of interest were then entered into a group-level model. Activated brain regions from this analysis are reported at p < .05, FDR corrected for multiple comparisons.

Experiment 2

Participants

Twenty right-handed healthy young participants with no red–green color blindness participated in this study (12 women, mean age = 25.5 years, SD = 4.85 years). All participants were naive with respect to the experimental hypothesis and were only informed of the required experimental task. The local ethics committee of the medical faculty at the Ludwig–Maximilian University approved the study. Informed written consent was obtained from all participants in accordance with the Declaration of Helsinki. One participant was excluded because of an anatomical abnormality for a final cohort of 19.

Stimuli and Design

In the second experiment, 8 of the 12 turning angles were used because no effect of turning angle was found. The difference in frame-wise local luminance changes from the first experiment was removed by reducing the overall contrast in indistinct stimuli. Therefore, if the same activity is found in response to these stimuli, then it is not related to differences in local high-frequency light intensity changes in the stimulus. With these stimuli, participants performed a detection task, which, in contrast to the first task, did not require participants to process the original content of the stimuli. Subjects indicated the presence of a barely visible red arrow, which appeared for 10 frames (0.166 sec) within both types of stimuli. The red component of each pixel was increased by 10/255, and the green and blue component was decreased such that the luminance of each pixel remained constant (Figure 1B and C). The arrows had one of four different lengths, could point either to the left or to the right, and were positioned in the middle of one of the four corner quadrants of an imaginary 3 × 3 grid (nine equal squares). Fifty percent of the stimuli contained an arrow, with an equal distribution of arrows appearing in tunnel and phase-scrambled stimuli distributed over 0.5–5.5 sec of the film. In addition, participants also performed the direction task from the first experiment to ensure that task-specific effects were not because of the new participant cohort.

Experimental Procedure

Participants were initially trained on both tasks to indicate via button press depending on task either (a) the presence of an arrow within the presented film or (b) the direction of optic flow motion (left or right). Subjects were preexposed to a single frame containing an arrow for both types of stimuli for 5 sec before training to increase the accuracy with which participants could perform the task. Detection performance was trained until 80% correct was reached. Subsequently, participants were trained on the direction task as they were trained in Experiment 1. Subjects were instructed to fixate a red fixation cross in the middle of the screen throughout training and during the experiment, and eye movements were monitored with an MRI-compatible camera with EyeSeeCam software (Schneider et al., 2009).

During the experiment, tasks were presented in a single block, with randomized task order across runs. Each run started with a six-scan (13.6 sec) fixation cross, followed by a three-scan (6.75 sec) task-specific instruction and then the task. The next task was separated by another six-scan fixation period and subsequent three-scan instruction period. Twenty-four stimuli (12 from each stimulus type, six blocks per stimulus) were presented for the direction task, and 48 stimuli (12 blocks per stimulus), for the arrow detection task. Within-task stimulus presentation rates were equalized across both runs, and each presented stimulus and arrow combination was unique. Visual stimuli were presented within a black frame with a projection system (60 Hz, screen resolution = 800 × 600, Christie LX40) and were viewed over a front surface mirror (field of view = 24° × 19°).

Data Acquisition

fMRI images were acquired using an eight-channel head coil on a 3-T whole-body MR scanner (Signa HDx) with a T2*-weighted gradient-echo, echo-planar sequence (repetition time = 2.25 sec, field of view = 220 mm, matrix = 64 × 64). Each volume consisted of 36 axial slices, each with a slice thickness of 3.5 mm with no interslice gap. Padding and adjustable head restraints were used to minimize head motion. A high-resolution T1-weighted anatomical image (0.8 × 0.8 × 0.8 mm isotropic voxels) was also acquired from each participant.

Data Analysis

Data were analyzed using SPM5 (Wellcome Trust, London, United Kingdom) for Matlab (MathWorks, Natick, MA). Data were preprocessed in the same way as in Experiment1. A high-pass filter (cutoff = 128 sec) was applied to removelow-frequency noise and slow signal drifts. Separate regressors for each stimulus and task combination as well as the behavioral responses modeled the BOLD time courses at the single-subject level. Six additional regressors modeled participant movement. A 2 × 2 factorial design including the interaction between both stimuli and task was used to model group-level effects. Subjects' and task-wise performance as a covariate were also entered into the model. Contrasts for main effects and interactions were analyzed using t test, thresholded at p < .05, FDR corrected for multiple comparisons. Behavioral performance was assessed with a 2 × 2 repeated-measures ANOVA with the factors Task (direction vs. detection) and Stimulus (tunnel vs. phase-scrambled) and was computed for percentage of correct responses.

PPI Analysis for Both Experiments

PPI analysis tests whether the neuronal responses in each voxel can be explained by the interaction between the neuronal activity in a given seed region (in this case, the hippocampus) and experiment-related cognitive processes, which, in this case, is viewing meaningful or indistinct stimuli (Friston et al., 1997). As PPI analysis typically involves a common seed region across participants, we created a bilateral ROI based on the hippocampal activation that was identified in the group analysis for indistinct stimuli of both experiments masked with an anatomical image of the posterior hippocampus (see Figure 4). The mean BOLD signal time course was extracted from this ROI for each participant and convolved with the canonical hemodynamic response function for the stimuli resulting in the interaction term. The mean BOLD signal, together with the regressor for each stimulus, and the interaction term were entered into a general linear model. Pearson's product–moment correlation coefficient was computed between the time course of every voxel in the brain and the interaction term. These correlation values were then converted to z values using Fisher's r-to-z transformation. To assess statistical significance across participants for each experiment, whole-brain voxelwise z maps were then entered into a group-level analysis where contrasts of interest were assessed with t tests thresholded at p < .05, FDR corrected for multiple comparisons.

RESULTS

Behavioral Analyses

The behavioral results of the first experiment were already published in Fraedrich et al. (2010) and showed a higher percentage of correct responses for meaningful tunnel stimuli. Both the difference in correct responses between stimulus types and the average performance from each individual, irrespective of stimulus, did not correlate with BOLD signal changes in the current analyses. In the second experiment, we tested for differences in performance between stimuli (meaningful vs. indistinct) and task (direction vs. detection). The 2 × 2 repeated-measures ANOVA revealed a main effect for Task, F(1, 18) = 14.1, p = .001, a main effect of Stimulus F(1, 18) = 8.8, p = .008, and a significant interaction of Task × Stimulus, F(1, 18) = 5.8, p = .027. Response accuracy for the detection task did not significantly differ for the meaningful and indistinct films (meaningful: 81.0% vs. indistinct: 80.1%). Performance during the detection task was worse than in the direction task, indicating that the detection task was more demanding. For the direction task, participants were more likely to recognize the correct optic flow direction for the meaningful stimuli (97.9%) than for the indistinct stimuli (84.5%), same as in Experiment 1. The behavioral measures were used to model hemodynamic effects in the group-level general linear model. Testing for the effect of performance did not reveal any significant correlations with brain activation.

For the direction task in the first experiment, the mean response times did not significantly differ [t(28) = 0.436, p = .667] for indistinct stimuli (mean = 638 msec, SD = 175 msec) and meaningful stimuli (mean = 630 msec, SD = 132 msec). For the second experiment, a repeated-measures ANOVA over response times (factors Stimulus and Task) revealed a significant main effect of Stimulus, F(1, 18) = 4.7, p = .043, a significant main effect for Task, F(1, 18) = 118.2, p < .001, and a significant interaction, F(1, 18) = 80.7, p < .001. The mean response time was lower in the direction task (indistinct: mean = 555 msec, SD = 40 msec; meaningful: mean = 527 msec, SD = 25 msec) compared with the detection task (indistinct: mean = 599 msec, SD = 57 msec; meaningful: mean = 619 msec, SD = 58 msec). The interaction is caused by shorter response times for meaningful stimuli than indistinct stimuli in the direction task, whereas in the detection task, response times were longer for meaningful stimuli.

Eye tracking data revealed that participants constantly fixated the target, and there were no differences between stimuli or conditions.

fMRI Analyses

Functional Connectivity of Early Visual Areas

Because early visual areas were active during presentation of indistinct stimuli, we only analyzed the functional connectivity of the early visual component extracted using ICA. The group analysis over the spatially normalized ROAs of early visual ICs showed significant activation not only in visual areas such as the right cuneus, left middle occipital area, and fusiform but also bilaterally in the hippocampus (Figure 2A), although the selection criteria for labeling this independent component was solely based on the activation in early visual areas. This implies a functional connectivity of early visual areas and the hippocampus that forms one independent BOLD process with highly correlated BOLD time courses. Thus, the viewing of the experimental stimuli evoked temporal dynamics found in the visual areas that can also be found in the posterior hippocampus.

Figure 2. 

Functional connectivity between the visual system and the hippocampus. (A) Axial brain slices (z = −12, −2, 24, 34; MRIcron anatomical template ch2bet) of the resulting thresholded group-level t statistic (p < .05, FDR corrected) from the individual early visual cortex independent component (n = 18). The test reveals regions that belong to this component in all participants that included visual areas and the hippocampus bilaterally, indicating a statistically independent process between these regions. (B, C) The average BOLD signal (i.e., back-projected component time course) for the component seen in A averaged over 18 participants, locked to the onset of indistinct (B) and meaningful (C) stimuli. The dotted vertical line indicates stimulus onset, and the shaded area represents standard deviation across participants. The time course shows consistently higher values during indistinct stimulus presentation and lower values for meaningful stimuli across all participants.

Figure 2. 

Functional connectivity between the visual system and the hippocampus. (A) Axial brain slices (z = −12, −2, 24, 34; MRIcron anatomical template ch2bet) of the resulting thresholded group-level t statistic (p < .05, FDR corrected) from the individual early visual cortex independent component (n = 18). The test reveals regions that belong to this component in all participants that included visual areas and the hippocampus bilaterally, indicating a statistically independent process between these regions. (B, C) The average BOLD signal (i.e., back-projected component time course) for the component seen in A averaged over 18 participants, locked to the onset of indistinct (B) and meaningful (C) stimuli. The dotted vertical line indicates stimulus onset, and the shaded area represents standard deviation across participants. The time course shows consistently higher values during indistinct stimulus presentation and lower values for meaningful stimuli across all participants.

The event-related BOLD responses computed for the independent component of early visual areas were found to be stimulus-locked to meaningful and indistinct visual motion stimuli. For indistinct stimuli, the back-projected component time course for early visual areas (corresponding to BOLD signal) showed an increase in BOLD signal, whereas the back-projected component time course in response to meaningful stimuli showed a relative decrease in signal strength (Figure 2B and C). This pattern of activation corresponds to the visual activation that has previously been found for these stimuli (Fraedrich et al., 2010). Therefore, the functional connectivity between the early visual areas and the posterior hippocampus was related to the stimuli.

Overall Stimulus-related Activity

For indistinct versus meaningful stimuli, the univariate analysis in the first experiment revealed significant activation in the occipital pole extending down to the lingual gyrus, the occipital fusiform gyrus, the hippocampus bilaterally, and the precentral and postcentral gyrus (Figure 3A). The differential hippocampal activation is also reflected in the percent signal change (Figure 3C). Consistent with previous findings that coherent shapes and meaningful objects lead to a reduction of primary visual cortex activity compared with scrambled low-level feature-matched counterparts (Murray, Kersten, Olshausen, Schrater, & Woods, 2002), our indistinct phase-scrambled stimuli lead to a strong increase in early visual cortex activation in comparison with the coherent and meaningful tunnel stimuli. Additionally, active regions were found in the insular cortex bilaterally, paracingulate gyrus, right frontal pole, superior frontal gyrus bilaterally, frontal orbital cortex bilaterally, left frontal pole, left inferior frontal gyrus, primary somatosensory cortex, and middle temporal cortex (see Figure 3A).

Figure 3. 

Hippocampal activity when viewing indistinct motion stimuli. (A) Activation map for indistinct stimuli from Experiment 1, the optic flow direction task (n = 29; p < .05, FDR corrected). (B) Activation map for indistinct stimuli from Experiment 2, the detection task (n = 19; p < .05, FDR corrected). Both tasks show mainly visual and right-dominant hippocampal activity, suggesting that the activity is stimulus-driven. Maps are superimposed on the MRIcron anatomic template ch2bet. (C, D) Percent signal change computed for Experiments 1 and 2 using the hippocampal group result of indistinct versus meaningful stimuli as ROI. After correction for the individual participant mean, percent signal change in response to both experimental tasks shows a positively increased signal change in the hippocampus in response to indistinct stimuli compared with meaningful ones. The difference in percent signal change for the direction task in Experiment 1 (C) and Experiment 2 (D, right) was not significant (ANOVA, Experiment × Stimulus, F(1, 46) = 1.93, p > .17).

Figure 3. 

Hippocampal activity when viewing indistinct motion stimuli. (A) Activation map for indistinct stimuli from Experiment 1, the optic flow direction task (n = 29; p < .05, FDR corrected). (B) Activation map for indistinct stimuli from Experiment 2, the detection task (n = 19; p < .05, FDR corrected). Both tasks show mainly visual and right-dominant hippocampal activity, suggesting that the activity is stimulus-driven. Maps are superimposed on the MRIcron anatomic template ch2bet. (C, D) Percent signal change computed for Experiments 1 and 2 using the hippocampal group result of indistinct versus meaningful stimuli as ROI. After correction for the individual participant mean, percent signal change in response to both experimental tasks shows a positively increased signal change in the hippocampus in response to indistinct stimuli compared with meaningful ones. The difference in percent signal change for the direction task in Experiment 1 (C) and Experiment 2 (D, right) was not significant (ANOVA, Experiment × Stimulus, F(1, 46) = 1.93, p > .17).

The hippocampus is known to be involved in classical conditioning that requires temporal integration over a delay period (Eichenbaum, Otto, & Cohen, 1994; Berger & Thompson, 1978). To correctly perform the direction task for the indistinct stimuli in the first experiment, participants had to integrate the direction of optic flow motion over time. Therefore, the direction task could have recruited memory processes, thus evoking hippocampal activation. For the meaningful stimuli, in contrast, it is not necessary to integrate information over the length of the stimulus, as the direction of the tunnel can be determined by processing the optic flow from a very short period of the stimulus. Another possible explanation for the differential hippocampal activity is the higher local luminance changes for indistinct stimuli, which could stimulate the hippocampus, similar to optic flow (Watrous, Fried, & Ekstrom, 2011).

The second experiment was used to control for these two possible effects. In addition to having participants respond to the direction of optic flow, participants were also asked to perform a detection task using the same stimuli because detection tasks do not require temporal integration and should not elicit hippocampal activation (Novitskiy et al., 2011; Hahn et al., 2009; Linden et al., 1999). The phase-scrambled stimuli were modified such that the local luminance changes were equal to their respective tunnel stimuli. In the second experiment, the detection task for indistinct versus meaningful stimuli led to activation in the occipital pole, precentral and postcentral gyrus, primary motor cortex, bilaterally inferior LOC (most likely corresponding to V5), and the hippocampus (Figure 3B; for hippocampal percent signal change, see Figure 3D).

No significant interaction between task and stimulus effects was found in the hemodynamic response. Therefore, we looked at the main effect of stimulus over both tasks, as well as the main effect of task. The main effect of stimulus type revealed activity for indistinct stimuli (Table 1) in the same areas found in the previous analysis. Despite matched local luminance changes between both film types and a globally reduced contrast for indistinct stimuli in the second experiment, the visual activation was still more pronounced for the indistinct stimuli in both task conditions. This corresponds to findings that V1 responses are increased to incoherent than to coherent motion and for less well-predictable motion (e.g., Bartels, Zeki, & Logothetis, 2008; Braddick et al., 2001; McKeefry, Watson, Frackowiak, Fong, & Zeki, 1997). The reduced activity in early visual areas during meaningful stimulus presentation might be based on predictive coding mechanisms (Rao & Ballard, 1999). Although participants were not required to perform temporal integration over time nor a detailed analysis of the spatio-temporal stimulus in the detection task, hippocampal activation was still found in response to indistinct stimuli. Task difficulty was matched between stimulus types for the detection task, and no significant correlations between BOLD signal change and behavioral response times or performance accuracy were found, suggesting that performance effects cannot explain the hippocampal activation found.

Table 1. 

Regions of Activity for Both Tasks from Experiment 2

Region
Hemisphere
Cluster Size (voxels)
Max. t
x, y, z (mm)
Indistinct > Meaningful (p < .05, FDR Corrected) 
Calcerine 10,008 15.91 0, −92, −10 
14.54 10, −98, −2 
13.98 26, −92, 2 
Postcentral 750 5.24 44, −28, 58 
4.87 52, −18, 48 
Precentral 4.76 42, −16, 58 
Postcentral 315 4.67 −54, −6, 44 
4.27 −58, −10, 38 
4.13 −50, −16, 52 
Frontal inf. operculum 398 4.04 38, 18, 18 
Frontal inf. tri 3.69 44, 14, 26 
Frontal inferior orbital 201 3.95 42, 32, −2 
2.98 30, 32, −6 
Mid frontal 245 3.54 36, 36, 18 
3.45 40, 44, 22 
2.93 38, 56, 16 
Hippocampus 32 3.49 24, −30, −6 
Primary motor cortex 165 3.23 −24, −24, 52 
Paracentral 3.21 −6, −30, 56 
Precentral 3.17 −28, −22, 66 
Superior temporal 35 3.15 58, −22, −2 
13 3.13 −58, −12, −2 
Insula 16 3.08 −30, −24, 8 
2.93 32, −24, 4 
Hippocampus 2.89 −24, −32, −6 
SMA 2.87 0, 6, 60 
 
Meaningful > Indistinct (p < .05, FDR Corrected) 
Temp. occ. fusiform 9,567 9.89 26, −44, −16 
Precuneus 9.49 14, −46, 46 
Fusiform 8.81 −26, −48, −10 
Middle occipital 666 5.98 −40, −80, 18 
Supramarginal 79 3.60 60, −30, 26 
Superior temporal 16 3.50 66, −42, 16 
Superior frontal 28 3.24 22, −2, 48 
Posterior cingulate 2.98 16, −38, 14 
Region
Hemisphere
Cluster Size (voxels)
Max. t
x, y, z (mm)
Indistinct > Meaningful (p < .05, FDR Corrected) 
Calcerine 10,008 15.91 0, −92, −10 
14.54 10, −98, −2 
13.98 26, −92, 2 
Postcentral 750 5.24 44, −28, 58 
4.87 52, −18, 48 
Precentral 4.76 42, −16, 58 
Postcentral 315 4.67 −54, −6, 44 
4.27 −58, −10, 38 
4.13 −50, −16, 52 
Frontal inf. operculum 398 4.04 38, 18, 18 
Frontal inf. tri 3.69 44, 14, 26 
Frontal inferior orbital 201 3.95 42, 32, −2 
2.98 30, 32, −6 
Mid frontal 245 3.54 36, 36, 18 
3.45 40, 44, 22 
2.93 38, 56, 16 
Hippocampus 32 3.49 24, −30, −6 
Primary motor cortex 165 3.23 −24, −24, 52 
Paracentral 3.21 −6, −30, 56 
Precentral 3.17 −28, −22, 66 
Superior temporal 35 3.15 58, −22, −2 
13 3.13 −58, −12, −2 
Insula 16 3.08 −30, −24, 8 
2.93 32, −24, 4 
Hippocampus 2.89 −24, −32, −6 
SMA 2.87 0, 6, 60 
 
Meaningful > Indistinct (p < .05, FDR Corrected) 
Temp. occ. fusiform 9,567 9.89 26, −44, −16 
Precuneus 9.49 14, −46, 46 
Fusiform 8.81 −26, −48, −10 
Middle occipital 666 5.98 −40, −80, 18 
Supramarginal 79 3.60 60, −30, 26 
Superior temporal 16 3.50 66, −42, 16 
Superior frontal 28 3.24 22, −2, 48 
Posterior cingulate 2.98 16, −38, 14 

Complete list of regions resulting from the second experiment. Montreal Neurological Institute coordinates of the peak voxel, t values, and cluster sizes (in number of voxels). L = left hemisphere; R = right hemisphere; inf = inferior; Temp. occ. = temporal occipital.

Task-specific Activity

We also tested for task-specific effects in the second experiment, independent of stimulus type. The contrast detection versus direction revealed no significant activation, whereas the opposite contrast yielded significant activation in bilateral frontal inferior gyrus (including Broca's area), inferior orbito-frontal, supramarginal gyrus (including Wernicke's region), inferior temporal gyrus, inferior parietal, superior frontal, precentral gyrus, optic radiation, and inferior occipital cortex (p < .05, FDR corrected). The language processing areas showed a left-hemispheric dominance and were most likely related to the linguistic nature of the response (“left” or “right”). The motion-sensitive area V5 was more active for the direction task, likely because of the relevance of visual motion for the task. Eye movements cannot account for the relative differences found in V5 because participants did not differ in their eye movements between tasks.

Connectivity Assessed with PPI Analysis

The PPI analysis disambiguates correlations of a spurious sort from those mediated by direct or indirect neuronal interactions. If the hippocampus is involved in processing indistinct stimuli, then not only the activity of the hippocampus but also its coupling is expected to be stimulus-dependent. We chose to assess this connectivity with a PPI analysis instead of ICA because PPI analyses inherently look at stimulus-related effects, and finding a hippocampal visual connectivity using PPI would suggest that the connectivity between these two regions is dependent on the stimulus. Both experiments were analyzed separately by calculating the within-subject correlation between the activity in the posterior hippocampus and activity in the rest of the brain. The two experiments did not differ in the connectivity patterns found, so we combined the PPI analysis over both experiments (n = 48). The posterior hippocampus showed stimulus-dependent correlations with the inferior temporal gyrus (specifically the temporal occipital fusiform gyrus), superior LOC (including 7a), inferior LOC or V5, and in inferior and superior parietal cortex, including the left supramarginal gyrus, for indistinct stimuli (Figure 4A, yellow). These cortical regions lie within the dorsal and ventral visual stream, associated with visuospatial or motion processing and object recognition, respectively. Thus, the hippocampus shows stimulus-specific connectivity to ventral and dorsal visual stream areas for these stimuli. In addition to these areas, the connectivity result also revealed activity in the bilateral posterior cingulate gyrus, an area that shows strong connectivity with the caudal inferior parietal lobe (Kravitz, Saleem, Baker, & Mishkin, 2011). For meaningful stimuli, the posterior hippocampus showed a stronger correlation with the occipital fusiform gyrus, including activity in early visual areas, in V5 bilaterally, in anterior insular cortex bilaterally, and in left precentral gyrus, as well as minor activation in paracentral gyrus bilaterally and right inferior frontal gyrus (Figure 4A, blue).

Figure 4. 

Stimulus-dependent functional connectivity of the posterior hippocampus with other cortical areas for indistinct and meaningful (Mngf) stimuli. (A) The posterior hippocampus showed an increased connectivity to areas along the dorsal and ventral visual stream in both experiments, when participants viewed indistinct stimuli (yellow). While viewing meaningful motion stimuli, the hippocampus showed an increased correlation with activity in early visual areas (blue). The seed region that was identical for both experiments is shown in red. Significantly correlated regions with the hippocampus are depicted on axial (z = −10, −5, 14, 58) slices to show the exact location of activity. The statistical threshold for both experiments was p < .05 (FDR corrected), and the anatomical template was the same as in Figure 2. (B, C) Correlations between hippocampal activity (x = 22, y = −30, z = −6) and activity from an exemplary preprocessed (detrended, high-pass filtered) voxel for one participant within the right occipital cortex (B; x = 16, y = −88, z = −16) or the right precuneus (C; x = 14, y = −54, z = 50) illustrate the found PPI for both stimulus types. The values on the x and y axes have arbitrary units.

Figure 4. 

Stimulus-dependent functional connectivity of the posterior hippocampus with other cortical areas for indistinct and meaningful (Mngf) stimuli. (A) The posterior hippocampus showed an increased connectivity to areas along the dorsal and ventral visual stream in both experiments, when participants viewed indistinct stimuli (yellow). While viewing meaningful motion stimuli, the hippocampus showed an increased correlation with activity in early visual areas (blue). The seed region that was identical for both experiments is shown in red. Significantly correlated regions with the hippocampus are depicted on axial (z = −10, −5, 14, 58) slices to show the exact location of activity. The statistical threshold for both experiments was p < .05 (FDR corrected), and the anatomical template was the same as in Figure 2. (B, C) Correlations between hippocampal activity (x = 22, y = −30, z = −6) and activity from an exemplary preprocessed (detrended, high-pass filtered) voxel for one participant within the right occipital cortex (B; x = 16, y = −88, z = −16) or the right precuneus (C; x = 14, y = −54, z = 50) illustrate the found PPI for both stimulus types. The values on the x and y axes have arbitrary units.

Some of the visual regions found in the PPI analysis overlapped with the areas that showed stimulus-dependent activity in the univariate analysis. To shed light on the connectivity patterns, we looked at the correlations in activity in an example voxel from the occipital cortex, the precuneus, and the hippocampus in a single participant. The voxel in the occipital cortex showed higher activation for the indistinct stimuli in accordance to the univariate results described above. It also showed correlated activity with the hippocampus for both stimulus types, with a slightly higher correlation for meaningful stimuli (Figure 4B). The precuneus showed a strong positive correlation to hippocampal activity for indistinct stimuli but not for meaningful tunnel stimuli despite higher overall activation of the precuneus for meaningful stimuli (Figure 4C). These differential effects suggest a hippocampal coupling with the precuneus dependent on perceptual inconsistency or constancy of the stimulus and a stable connectivity between early visual areas and the hippocampus (consistent with the ICA results).

DISCUSSION

Using complementary analysis techniques in two fMRI experiments, we examined the hippocampal recruitment to meaningful visual motion stimuli and the corresponding phase-scrambled indistinct stimuli as well as the functional connectivity between visual and hippocampal areas. Without an explicit memory task, we consistently observed visual and posterior bilateral hippocampal activation in response to indistinct visual motion stimuli. The results of our two experiments showed that hippocampal activation was independent of image statistics and task. The activation was also found for indistinct stimuli in the detection task, for which an explicit processing of the stimulus content was not required. This indicates that the hippocampal activation is related to implicit processing of the stimulus. Furthermore, the ICA revealed a stimulus-related functional connectivity between the visual cortex and the hippocampus, reflecting mnemonic information processing based on visual sensory input. In addition, the coupling revealed by PPI analysis between the posterior hippocampus and areas within ventral and dorsal visual stream, encompassing the inferior LOC (most likely V5) and the superior parietal cortex bilaterally, was also independent of experiment.

Hippocampal Recruitment for Indistinct Visual Motion Stimuli

Participants acquired a mental representation for both stimulus types because they were exposed to both stimuli during training and could give a general description of their appearance when asked. Therefore, although novelty has been associated with hippocampal activation (Kumaran & Maguire, 2007; Nyberg, 2005), stimulus novelty in the sense that a stimulus has not been experienced before cannot explain our hippocampal results as participants had equal exposure to both stimulus types. Other aspects of novelty such as associative and contextual novelty also cannot explain our present findings (cf. Kumaran & Maguire, 2006). Our stimuli did not allow participants to develop an association that could have then been violated, and the context did not change during the experiment. Instead, we believe that the temporally uncertain nature of the stimuli, which can also be seen as a form of novelty, was critical for hippocampal recruitment.

Although the phase manipulation did not change the spatio-temporal amplitude spectrum between both stimulus types, the phase manipulation in the temporal dimension introduced structural changes over time. Previous studies presenting static images with manipulated phase information have not reported hippocampal activation (Wichmann et al., 2006; Olman et al., 2004), suggesting that the structural changes introduced through phase manipulation in the temporal dimension of our stimuli are the determining factor for hippocampal recruitment. Because both stimulus types were matched for image statistics and optic flow and, in the second experiment, also matched for local luminance changes, these factors cannot account for the differential hippocampal activation. The causal difference seems to lie in the differential structural information over time between both stimulus types.

The meaningful stimuli possess a clear and temporally very consistent structure with only minor structural changes over time, which is represented by the significantly higher mutual information between frames. The decreased visual activation in response to meaningful stimuli might speak for predictive coding mechanisms taking place. In contrast, the indistinct stimuli have a continuously changing structure and thus lead to a continuous change in current sensory input. Indistinct stimuli contained less mutual information between frames. Thus, for meaningful stimuli, uncertainty about the following frame is reduced by knowing the present one, whereas indistinct stimuli still defy expectation and thus could drive memory encoding processes to a greater degree. The hippocampal activity was found irrespective of the task participants performed, and the detection task did not explicitly require the processing of the stimulus, suggesting that the hippocampus implicitly processes unpredictable dynamic stimuli.

Stimulus-dependent Hippocampal Connectivity

In both experiments, the same coupling between the hippocampus and areas in the ventral and dorsal visual stream was found for indistinct stimuli. With regard to the roles of these visual streams, this suggests a functional relationship between the hippocampus and object and place recognition centers for these particular stimuli. The connectivity was consistent despite different task demands. The ventral visual stream, in particular, the lateral–occipital complex, is a hub for object recognition (Malach et al., 1995), and activation in this region correlates highly with recognition performance (Grill-Spector, Kushnir, Hendler, & Malach, 2000). Furthermore, activation in bilateral occipito-temporal areas (corresponding to region LO) has been found in response to perceptual closure processes that enable recognition despite only partial visual information (Doniger et al., 2000).

The connections to the dorsal visual stream encompassed regions from the general occipito-parietal system that is known for visuospatial and motion processing (Kravitz et al., 2011; Born & Bradley, 2005). Furthermore, we found a projection from the hippocampus to the inferior parietal cortex that has previously been described as part of the parieto-medial pathway in monkeys (Kravitz et al., 2011). This pathway is implicated in optic flow processing (Phinney & Siegel, 2000); however, this alone cannot explain our results because optic flow and local luminance changes were identical for both stimulus types. Despite the task difference for both experiments and the inclusion of a colored stimulus in the detection task, it is intriguing that the same hippocampal–cortical connectivity pattern was observed across experiments.

Our functional connectivity results are in agreement with electrophysiological and histological findings in rats that the hippocampus receives crucial sensory input from the visual cortex and that the dorsal visual cortex projects multisynaptically via occipital connections and the ventral visual pathway, via temporal connections to the hippocampus (Tsanov & Manahan-Vaughan, 2008). Also, in primates, a direct connection between parietal (area 7a and b) and temporal regions to the hippocampus has been found (Rockland & Van Hoesen, 1999). Furthermore, we confirm earlier observations of distinct intrinsic functional connectivity in humans between the posterior hippocampus and the parietal cortex (Kahn, Andrews-Hanna, Vincent, Snyder, & Buckner, 2008). The fact that the two complementary analyses, independent component and univariate analysis, associated the posterior hippocampus with other visual processing systems corresponds to the cytological and molecular boundaries of the hippocampus (Fanselow & Dong, 2010) and to the proposed role of the posterior hippocampus in visual–spatial processing (Hüfner, Strupp, Brandt, Smith, & Jahn, 2011; Maguire et al., 2000).

In response to meaningful stimuli, the connectivity analysis mainly revealed a coupling of the posterior hippocampus to early visual areas as well as minor activations to other cortical areas such as bilateral paracingulate gyrus and anterior insular cortex. It is important to note that the hippocampus was not active in response to the meaningful stimuli, thus the connectivity pattern found represents a fundamental connectivity between the hippocampus and visual cortex that is independent of experiment.

Unpredictability as Determining Factor for Hippocampal Involvement?

Indistinct stimuli have less mutual information between frames (see Methods). Thus, based on current sensory input, the future visual input is less predictable and possibly causes a mismatch between the actual and predicted stimulus presentation leading to either a continual updating of the mental representation or learning to expect uncertainty, both reliant on the hippocampus. This concept would be akin to evidence linking hippocampal activation to higher unpredictability of visual stimuli (Strange et al., 2005) and, as such, complements the broad literature on the role of the hippocampus in novelty detection. Furthermore, the finding that the hippocampus seems to be especially active when visual input changes dynamically and unpredictably and thus mismatches prior expectations expands the view of the hippocampus as a continuous integrator of new information through updating of mental representations (McKenzie & Eichenbaum, 2011).

Given that forward models of prediction have been suggested for the hippocampus (Schacter & Addis, 2009) as well as for motion pattern analysis of the visual system through behavioral work (Roach, McGraw, & Johnston, 2011), the stimulus-dependent connectivity result between the hippocampus and visual processing areas could bridge the gap between these two separate findings.

From a system dynamics perspective that views the hippocampus as an attractor network (Rolls, 2007), the hippocampal activation could be explained by the combination of continuous structural change with the naturalistic motion still contained in the indistinct stimuli that keep the activation in the hippocampus in a reverberating state, which does not easily converge onto a stable point within the hippocampal attractor network. According to the proposal that vision can be thought of as “recognition-by-analogy” by which the visual input is linked to existing information stored in analogous memory representations (Bar, 2007, 2009), the task independence of the hippocampal activation while viewing indistinct visual motion stimuli with the cortical coupling to visuospatial and object recognition areas may represent the neuronal substrate for the attempt to combine information from both streams to recognize these indistinct stimuli.

Conclusion

Taken together, these functional data demonstrate that the hippocampus is recruited in response to indistinct visual motion stimuli with a temporally unpredictable nature through an interaction between perceptual and mnemonic processes. The pattern of cortico-hippocampal connectivity, in the absence of an explicit memory task, provides evidence for the hippocampus in binding neocortical visual processing areas. Overall, our present findings demonstrate that higher cognitive areas are recruited in response to purely visual tasks and that functional cortico-hippocampal connectivity is flexible and changes depending on perceptual demands.

Acknowledgments

This study was supported by BMBF (BCCN 01GQ0440, IFB 01EO0901), DFG (GRK 1091), Graduate School of Systemic Neurosciences (GSC 82/1), Taiwan Department of Health Clinical Trial and Research Center of Excellence (DOH99-TDB111-004).

Reprint requests should be sent to Eva Fraedrich, Neurologisches Forschungshaus, LMU, Marchioninistrasse 23, 81377 München, Germany, or via e-mail: eva.fraedrich@lrz.uni-muenchen.de.

Amaral
,
D. G.
, &
Witter
,
M. P.
(
1989
).
The three dimensional organization of the hippocampal formation: A review of anatomical data
.
Neuroscience
,
31
,
571
591
.
Bar
,
M.
(
2007
).
The proactive brain: Using analogies and associations to generate predictions
.
Trends in Cognitive Sciences
,
11
,
280
289
.
Bar
,
M.
(
2009
).
The proactive brain: Memory for predictions
.
Philosophical Transactions of the Royal Society of London, Series B, Biological Sciences
,
364
,
1235
1243
.
Barense
,
M. D.
,
Henson
,
R. N. A.
, &
Graham
,
K. S.
(
2011
).
Perception and conception: Temporal lobe activity during complex discriminations of familiar and novel faces and objects
.
Journal of Cognitive Neuroscience
,
23
,
3052
3067
.
Bartels
,
A.
,
Zeki
,
S.
, &
Logothetis
,
N.
(
2008
).
Natural vision reveals regional specialization to local motion and to contrast-invariant, global flow in the human brain
.
Cerebral Cortex
,
18
,
705
717
.
Bell
,
A. J.
, &
Sejnowski
,
T. J.
(
1995
).
An information-maximization approach in blind separation and blind deconvolution
.
Neural Computation
,
7
,
1004
1034
.
Berger
,
T. W.
, &
Thompson
,
R. F.
(
1978
).
Neuronal plasticity in the limbic system during classical conditioning of the rabbit nictitating membrane response: I. The hippocampus
.
Brain Research
,
145
,
323
346
.
Binder
,
J. R.
,
Bellgowan
,
P. S.
,
Hammeke
,
T. A.
,
Possing
,
E. T.
, &
Frost
,
J. A.
(
2005
).
A comparison of two fMRI protocols for eliciting hippocampal activation
.
Epilepsia
,
46
,
1061
1070
.
Bird
,
C. M.
, &
Burgess
,
N.
(
2008
).
The hippocampus and memory: Insights from spatial processing
.
Nature Reviews Neuroscience
,
9
,
182
194
.
Born
,
R. T.
, &
Bradley
,
D. C.
(
2005
).
Structure and function of visual area MT
.
Annual Reviews Neuroscience
,
28
,
157
189
.
Braddick
,
O. J.
,
O'Brien
,
J. M. D.
,
Wattam-Bell
,
J.
,
Atkinson
,
J.
,
Hartley
,
T.
, &
Turner
,
R.
(
2001
).
Brain areas sensitive to coherent visual motion
.
Perception
,
30
,
61
72
.
Buckner
,
R. L.
(
2010
).
The role of the hippocampus in prediction and imagination
.
Annual Review of Psychology
,
61
,
27
48
.
Burwell
,
R. D.
, &
Agster
,
K. L.
(
2008
).
Anatomy of the hippocampus and the declarative memory system
.
In H. Eichenbaum (Ed.)
,
Memory systems, Vol. [3] of Learning and memory: A comprehensive reference, 4 vols. (J. Byrne, Ed.)
(pp.
47
66
).
Oxford
:
Elsevier
.
Carr
,
V. A.
,
Rissman
,
J.
, &
Wagner
,
A. D.
(
2010
).
Imaging the human medial temporal lobe with high resolution fMRI
.
Neuron
,
65
,
298
308
.
Doniger
,
G. M.
,
Foxe
,
J. J.
,
Murray
,
M. M.
,
Higgins
,
B. A.
,
Snodgrass
,
J. G.
,
Schroeder
,
C. E.
,
et al
(
2000
).
Activation timecourse of ventral visual stream object-recognition areas: High density electrical mapping of perceptual closure processes
.
Journal of Cognitive Neuroscience
,
2
,
615
621
.
Duann
,
J. R.
,
Jung
,
T. P.
,
Kuo
,
W. J.
,
Yeh
,
T. C.
,
Makeig
,
S.
,
Hsieh
,
J. C.
,
et al
(
2002
).
Single trial variability in event-related BOLD signals
.
NeuroReport
,
15
,
823
835
.
Duncan
,
K.
,
Ketz
,
N.
,
Inati
,
S. J.
, &
Davachi
,
L.
(
2012
).
Evidence for CA1 as a match/mismatch detector: A high resolution fMRI study of the human hippocampus
.
Hippocampus
,
22
,
389
398
.
Eichenbaum
,
H.
(
2004
).
Hippocampus: Cognitive processes and neural representations that underlie declarative memory
.
Neuron
,
44
,
109
120
.
Eichenbaum
,
H.
,
Otto
,
T.
, &
Cohen
,
N. J.
(
1994
).
Two functional components of the hippocampal memory system
.
Behavioral and Brain Sciences
,
17
,
449
472
.
Eldridge
,
L. L.
,
Engel
,
S. A.
,
Zeineh
,
M. M.
,
Bookheimer
,
S. Y.
, &
Knowlton
,
B. J.
(
2005
).
A dissociation of encoding and retrieval processes in the human hippocampus
.
Journal of Neuroscience
,
25
,
3280
3286
.
Fanselow
,
M. S.
, &
Dong
,
H.-W.
(
2010
).
Are the dorsal and ventral hippocampus functionally distinct structures?
.
Neuron
,
65
,
7
19
.
Fraedrich
,
E. M.
,
Glasauer
,
S.
, &
Flanagin
,
V. L.
(
2010
).
Spatiotemporal phase-scrambling increases visual cortex activity
.
NeuroReport
,
21
,
596
600
.
Friston
,
K. J.
,
Buechel
,
C.
,
Fink
,
G. R.
,
Morris
,
J.
,
Rolls
,
E.
, &
Dolan
,
R. J.
(
1997
).
Psychophysiological and modulatory interactions in neuroimaging
.
Neuroimage
,
6
,
218
229
.
Greicius
,
M. D.
, &
Menon
,
V.
(
2004
).
Default-mode activity during a passive sensory task: Uncoupled from deactivation but impacting activation
.
Journal of Cognitive Neuroscience
,
16
,
1484
1492
.
Grill-Spector
,
K.
,
Kushnir
,
T.
,
Hendler
,
T.
, &
Malach
,
R.
(
2000
).
The dynamics of object-selective activation correlate with recognition performance in humans
.
Nature Neuroscience
,
3
,
837
843
.
Hahn
,
B.
,
Ross
,
T. J.
,
Wolkenberg
,
F. A.
,
Shakleya
,
D. M.
,
Huestis
,
M. A.
, &
Stein
,
E. A.
(
2009
).
Performance effects of nicotine during selective attention, divided attention, and simple stimulus detection: An fMRI study
.
Cerebral Cortex
,
19
,
1990
2000
.
Hasselmo
,
M. E.
,
Schnell
,
E.
, &
Barkai
,
E.
(
1995
).
Dynamics of learning and recall at excitatory recurrent synapses and cholinergic modulation in rat hippocampal region CA3
.
The Journal of Neuroscience
,
15
,
5249
5262
.
Hüfner
,
K.
,
Strupp
,
M.
,
Brandt
,
T.
,
Smith
,
P.
, &
Jahn
,
K.
(
2011
).
Spatial separation of visual and vestibular processing in the human hippocampal formation
.
Annals of the New York Academy of Sciences
,
1233
,
177
186
.
Kahn
,
I.
,
Andrews-Hanna
,
J. R.
,
Vincent
,
J. L.
,
Snyder
,
A. Z.
, &
Buckner
,
R. L.
(
2008
).
Distinct cortical anatomy linked to subregions of the medial temporal lobe revealed by intrinsic functional connectivity
.
Journal of Neurophysiology
,
100
,
129
139
.
Kravitz
,
D. J.
,
Saleem
,
K. S.
,
Baker
,
C. I.
, &
Mishkin
,
M.
(
2011
).
A new neural framework for visuospatial processing
.
Nature Reviews Neuroscience
,
12
,
217
230
.
Kumaran
,
D.
, &
Maguire
,
E. A.
(
2006
).
An unexpected sequence of events: Mismatch detection in the human hippocampus
.
PLoS Biology
,
4
,
2372
2382
.
Kumaran
,
D.
, &
Maguire
,
E. A.
(
2007
).
Which computational mechanisms operate in the hippocampus during novelty detection?
.
Hippocampus
,
17
,
735
748
.
Lavenex
,
P.
, &
Amaral
,
D. G.
(
2000
).
Hippocampal–neocortical interaction: A hierarchy of associativity
.
Hippocampus
,
10
,
420
430
.
Levy
,
W. B.
(
1996
).
A sequence predicting CA3 is a flexible associator that learns and uses context to solve hippocampal-like tasks
.
Hippocampus
,
6
,
579
590
.
Li
,
N.
, &
DiCarlo
,
J. J.
(
2010
).
Unsupervised natural visual experience rapidly reshapes size-invariant object representation in inferior temporal cortex
.
Neuron
,
67
,
1062
1075
.
Linden
,
D. E.
,
Prvulovic
,
D.
,
Formisano
,
E.
,
Völlinger
,
M.
,
Zanella
,
F. E.
,
Goebel
,
R.
,
et al
(
1999
).
The functional neuroanatomy of target detection: An fMRI study of visual and auditory oddball tasks
.
Cerebral Cortex
,
9
,
815
823
.
Lisman
,
J. E.
, &
Grace
,
A. A.
(
2005
).
The hippocampal-VTA loop: Controlling the entry of information into long-term memory
.
Neuron
,
46
,
703
713
.
Maguire
,
E. A.
,
Gadian
,
D. G.
,
Johnsrude
,
I. S.
,
Good
,
C. D.
,
Ashburner
,
J.
,
Frackowiak
,
R. S. J.
,
et al
(
2000
).
Navigation-related structural change in the hippocampi of taxi drivers
.
Proceedings of the National Academy of Sciences, U.S.A.
,
97
,
4398
4403
.
Makeig
,
S.
,
Bell
,
A. J.
,
Jung
,
T.-P.
, &
Sejnowski
,
T. J.
(
1996
).
Independent component analysis of electroencephalographic data
.
In D. Touretzky, M. Mozer, & M. Hasselmo (Eds.)
,
Advances in neural information processing systems
(pp.
145
151
).
Cambridge, MA
:
MIT Press
.
Malach
,
R.
,
Reppas
,
J. B.
,
Benson
,
R. R.
,
Kwong
,
K. K.
,
Jiang
,
H.
,
Kennedy
,
W. A.
,
et al
(
1995
).
Object-related activity revealed by functional magnetic resonance imaging in human occipital cortex
.
Proceedings of the National Academy of Sciences, U.S.A.
,
92
,
8135
8139
.
Martin
,
A.
(
1999
).
Automatic activation of the medial temporal lobe during encoding: Lateralized influences of meaning and novelty
.
Hippocampus
,
9
,
62
70
.
McKeefry
,
D. J.
,
Watson
,
J. D. G.
,
Frackowiak
,
R. S. J.
,
Fong
,
K.
, &
Zeki
,
S.
(
1997
).
The activity in human areas V1/V2, V3 and V5 during the perception of coherent and incoherent motion
.
Neuroimage
,
5
,
1
12
.
McKenzie
,
S.
, &
Eichenbaum
,
H.
(
2011
).
Consolidation and reconsolidation: Two lives of memories?
.
Neuron
,
71
,
224
233
.
McKeown
,
M. J.
,
Makeig
,
S.
,
Brown
,
G. G.
,
Jung
,
T. P.
,
Kindermann
,
S. S.
,
Bell
,
A. J.
et al
(
1998
).
Analysis of fMRI data by blind separation into independent spatial components
.
Human Brain Mapping
,
6
,
160
188
.
Murray
,
S. O.
,
Kersten
,
D.
,
Olshausen
,
B. A.
,
Schrater
,
P.
, &
Woods
,
D. L.
(
2002
).
Shape perception reduces activity in human primary visual cortex
.
Proceedings of the National Academy of Sciences, U.S.A.
,
99
,
15164
15169
.
Norman
,
K. A.
, &
O'Reilly
,
R. C.
(
2003
).
Modeling hippocampal and neocortical contributions to recognition memory: A complementary-learning-systems approach
.
Psychological Review
,
110
,
611
646
.
Novitskiy
,
N.
,
Ramautar
,
J. R.
,
Vanderperren
,
K.
,
De Vos
,
M.
,
Mennes
,
M.
,
Mijovic
,
B.
,
et al
(
2011
).
The BOLD correlates of the visual P1 and N1 in single-trial analysis of simultaneous EEG-fMRI recordings during a spatial detection task
.
Neuroimage
,
54
,
824
835
.
Nyberg
,
L.
(
5
).
Any novelty in hippocampal formation and memory?
.
Current Opinion in Neurology
,
18
,
424
428
.
Olman
,
C. A.
,
Ugurbil
,
K.
,
Schrater
,
P.
, &
Kersten
,
D.
(
2004
).
BOLD fMRI and psychophysiological measurements of contrast response to broadband images
.
Vision Research
,
44
,
669
683
.
Phinney
,
R. E.
, &
Siegel
,
R. M.
(
2000
).
Speed selectivity for optic flow in area 7a of the behaving macaque
.
Cerebral Cortex
,
10
,
413
421
.
Rao
,
R. P. N.
, &
Ballard
,
D. H.
(
1999
).
Predictive coding in the visual cortex: A functional interpretation of some extra-classical receptive-field effects
.
Nature
,
2
,
79
87
.
Roach
,
N. W.
,
McGraw
,
P. V.
, &
Johnston
,
A.
(
2011
).
Visual motion produces a forward prediction of spatial pattern
.
Current Biology
,
21
,
740
745
.
Rockland
,
K. S.
, &
Van Hoesen
,
G. W.
(
1999
).
Some temporal and parietal cortical connections converge in CA1 of the primate hippocampus
.
Cerebral Cortex
,
9
,
232
237
.
Rolls
,
E. T.
(
2007
).
An attractor network in the hippocampus: Theory and neurophysiology
.
Learning and Memory
,
14
,
714
731
.
Rudy
,
J. W.
, &
O'Reilly
,
R. C.
(
2001
).
Conjunctive representations, the hippocampus, and contextual fear conditioning
.
Cognitive, Affective, & Behavioral Neuroscience
,
1
,
66
82
.
Schacter
,
D. L.
, &
Addis
,
D. R.
(
2009
).
On the nature of medial temporal lobe contributions to the constructive simulation of future events
.
Philosophical Transactions of the Royal Society London, Series B, Biological Sciences
,
364
,
1245
1253
.
Schneider
,
E.
,
Villgrattner
,
T.
,
Vockeroth
,
J.
,
Bartl
,
K.
,
Kohlbecher
,
S.
,
Bardins
,
S.
,
et al
(
2009
).
EyeSeeCam: An eye movement-driven head camera for the examination of natural visual exploration
.
Annals of the New York Academy of Sciences
,
1164
,
461
467
.
Strange
,
B.
,
Duggins
,
A.
,
Penny
,
W.
,
Dolan
,
R. J.
, &
Friston
,
K. J.
(
2005
).
Information theory, novelty and hippocampal responses: Unpredicted or unpredictable?
.
Neural Networks
,
18
,
225
230
.
Tsanov
,
M.
, &
Manahan-Vaughan
,
D.
(
2008
).
Synaptic plasticity from visual cortex to hippocampus: Systems integration in spatial information processing
.
Neuroscientist
,
14
,
584
597
.
Van de Ven
,
V. G.
,
Formisano
,
E.
,
Prvulovic
,
D.
,
Roeder
,
C. H.
, &
Linden
,
D. E. J.
(
2004
).
Functional connectivity as revealed by spatial independent component analysis of fMRI measurements during rest
.
Human Brain Mapping
,
22
,
165
178
.
Watrous
,
A. J.
,
Fried
,
I.
, &
Ekstrom
,
A. D.
(
2011
).
Behavioral correlates of human hippocampal delta and theta oscillations during navigation
.
Journal of Neurophysiology
,
105
,
1747
1755
.
Wichmann
,
F. A.
,
Braun
,
D. I.
, &
Gegenfurtner
,
K. R.
(
2006
).
Phase noise and the classification of natural images
.
Vision Research
,
46
,
1520
1529
.