Abstract

Memory training (MT) in older adults with memory deficits often leads to frustration and, therefore, is usually not recommended. Here, we pursued an alternative approach and looked for transfer effects of 1-week attentional filter training (FT) on working memory performance and its neuronal correlates in young healthy humans. The FT effects were compared with pure MT, which lacked the necessity to filter out irrelevant information. Before and after training, all participants performed an fMRI experiment that included a combined task in which stimuli had to be both filtered based on color and stored in memory. We found that training induced processing changes by biasing either filtering or storage. FT induced larger transfer effects on the untrained cognitive function than MT. FT increased neuronal activity in frontal parts of the neuronal gatekeeper network, which is proposed to hinder irrelevant information from being unnecessarily stored in memory. MT decreased neuronal activity in the BG part of the gatekeeper network but enhanced activity in the parietal storage node. We take these findings as evidence that FT renders working memory more efficient by strengthening the BG–prefrontal gatekeeper network. MT, on the other hand, simply stimulates storage of any kind of information. These findings illustrate a tight connection between working memory and attention, and they may open up new avenues for ameliorating memory deficits in patients with cognitive impairments.

INTRODUCTION

The amount of information humans can store in their working memory (WM) is very much limited and varies considerably between individuals (Cowan, 2000). Effective WM performance requires attention to select relevant information. Hence, filtering out distractors is a basic requisite to encode, maintain, and manipulate relevant information in human WM (Luck & Vogel, 2013; Cowan & Morey, 2009; Vogel & Machizawa, 2004). In this respect, it has been shown that the individual WM capacity is correlated with the efficiency with which one filters our irrelevant information (Vogel, McCollough, & Machizawa, 2005). The filtering process has been associated with a neuronal gatekeeper network, which includes the BG and pFC and which controls access to WM storage and sensory processing in posterior brain regions (McNab & Klingberg, 2008; Ress, Backus, & Heeger, 2000). The storage process, on the other hand, has been attributed to the posterior parietal cortex (PPC), where neuronal activity levels mirror the number of items held in WM (Todd & Marois, 2005; Vogel et al., 2005). Persons with a high WM capacity because of their efficient filtering abilities, therefore, show not only high activity in the fronto-striatal gatekeeper network but also low distractor-related “unnecessary” parietal storage activity.

The proposed strong interaction between filtering efficiency and WM capacity made us ask whether a short, intensive training of filtering abilities can enhance WM performance by strengthening the gatekeeper network, which inhibits distracting information from entering memory. We compared the filtering training (FT) with a pure memory training (MT) program, which did not involve distractor inhibition. Twenty-nine young healthy participants were randomized to these two groups and underwent a 1-week training program. Longer, less intensive working MT protocols have shown that neuronal activation changes follow an inverted U-shape: Activation increases during the first 2 weeks but then starts to decrease as performance gains consolidate (Hempel et al., 2004). Hence, short training programs in which performance has not reached ceiling levels are more likely to reveal neuronal changes. Before and after the training, all participants underwent fMRI while they performed a task that combined the trained processes. This combined condition (WM+) was the one of interest for the training effects on brain function as it challenged both filtering and storage processes but was not part of any of the training protocols. This excluded that any change in neuronal activation was simply based on participants being more familiar with an often practiced task.

On the basis of the idea that filter abilities determine WM performance but not vice versa, we expected stronger transfer effects on the untrained function (i.e., far transfer) after FT. Concerning the neuronal correlates, we expected FT to induce activity changes in the frontal and MT to modulate activity in the parietal storage node.

METHODS

Participants and Procedure

A group of young, healthy right-handed participants (n = 15, mean age = 24.4 years, eight women) received the filter training (FT), and a second group (n = 14, mean age = 24.3 years, seven women) participated in the MT. Before and after the training, all participants underwent functional brain MR scanning while they performed a task that involved the two training conditions plus two extra conditions: an attention task that did not require filter processes and a memory task with filtering demands. To investigate the impact of both training regimes on brain function, we focused on the pre–post differences in the WM task with distractors as this condition constituted a hybrid of both functions under investigation.

Tasks and Stimuli

Pre- and Postmeasurement

Before and after training, all 29 participants performed an experimental task while lying in the MR scanner. The participants had to detect a change between two arrays, which were presented in serial (WM) or parallel (ATT) order. The change detection experiment included the following four subconditions: attention with distractors (ATT+), attention without distractors (ATT−), WM with distractors (WM+), and WM without distractors (WM−).

In the attention conditions, participants had to compare two stimuli arrays presented simultaneously next to each other, which were composed of a varying number of colored bars (see Figure 1A later in the paper). In ATT+ trials, they were instructed by a red or green cue to compare either only the red or green bars of the double array, while ignoring bars of the other color. In ATT− trials, the cue was black indicating that all bars had to be compared, which were then displayed in a single color, that is, there was no need to filter out irrelevant stimuli. Both attention conditions did not involve a memory component. In the WM conditions, participants had to compare two arrays presented consecutively with a delay of 900–1400 msec at the center of the screen. Again, in WM+ trials, a colored cue indicated which bars were relevant and which could be ignored based on their color, whereas in WM− trials, only bars of one color were shown after a black cue had been presented. Hence, the WM− condition lacked the necessity to filter out irrelevant distractors, whereas the WM+ combined both storage and filtering.

Figure 1. 

(A) Paradigms used during filter (FT, left) and memory (MT, right) training and for the assessment of training-induced brain activation changes (WM+, center). During FT, participants had to decide whether simultaneously presented bars in a predefined color matched in terms of orientation. Hence, FT trained attentional selection but not memory. During MT, two consecutively presented arrays without distractors had to be compared. MT, therefore, trained memory storage but not attentional selection. The WM+ condition involved both memory storage and attentional selection. Behavioral results of the two training regimens are presented for (B) attention trials, (C) memory trials, and (D) WM+ trials.

Figure 1. 

(A) Paradigms used during filter (FT, left) and memory (MT, right) training and for the assessment of training-induced brain activation changes (WM+, center). During FT, participants had to decide whether simultaneously presented bars in a predefined color matched in terms of orientation. Hence, FT trained attentional selection but not memory. During MT, two consecutively presented arrays without distractors had to be compared. MT, therefore, trained memory storage but not attentional selection. The WM+ condition involved both memory storage and attentional selection. Behavioral results of the two training regimens are presented for (B) attention trials, (C) memory trials, and (D) WM+ trials.

The no-distractor trials (WM−, ATT−) consisted of 4–6 horizontal and vertical bars (1.43° × 0.29°) of one color (either red or green). The other half of the trials contained distractors (WM+, ATT+) whereby an equal number of red and green bars were presented (i.e., 4–6 bars in the target and 4–6 in the distractor color per array). All bar arrays were presented within a 4° × 9.3° rectangular region against a gray background. In the ATT trials, the arrays were placed 1.79° to the right and left of the central fixation cross, and the cue was shaped as a double arrow. In the WM trials, the arrays were shown in the middle of the screen, and a square was used as cue. The participants' task was to decide whether the simultaneously (ATT) or consecutively (WM) presented arrays matched in terms of orientation of the bars in the relevant color. In half of the trials, the orientation of one target changed from vertical to horizontal or vice versa, and in the other half of the trials, no orientation change occurred. Every participant performed 72 trials of each of the four conditions. The different conditions were presented in blocks (runs) with a randomized order that was counterbalanced across participants and sessions.

Training

Four days after the initial MR scan, every participant underwent five training sessions, 1 hr a day (Monday–Friday). The FT participants (n = 15) received a distractor inhibition training without a memory component (see Figure 1A later in the paper); the MT group (n = 14; Figure 1C) underwent storage training without the need to filter out irrelevant distractors. Most of the 600 training trials (528) were especially created for the training sessions and were not part of the pre/post assessments. In every daily session, participants were presented with 200 training trials whereby the second and third/fourth and fifth days were composed of the same trials in a randomized order. To maximize the training effects, the trials became more difficult over the week by successively increasing the portion of larger set sizes (see Table 1 for details).

Table 1. 

Day Proportions of Trial Difficulty for ATT+ and WM Training in Percent

DifficultyDay 1Days 2 and 3Days 4 and 5
Four targets 50 25 25 
Five targets 25 50 25 
Six targets 25 25 50 
DifficultyDay 1Days 2 and 3Days 4 and 5
Four targets 50 25 25 
Five targets 25 50 25 
Six targets 25 25 50 

WM training consisted only of targets. In ATT+, the same number of distractors was added.

MRI Acquisition

Brain images were collected with a 3-T Siemens Magnetom Verio scanner (Erlangen, Germany) 4 days before the trainings started and 4 days after the trainings had been completed. T2*-weighted transverse, gradient echo, spiral echo-planar images were acquired with repetition time = 2000 msec, echo time = 30 msec, flip angle = 80°, 34 axial slices, 3.00-mm thickness, and 192 mm × 192 mm field of view resulting in voxel sizes of 2.0 × 2.0 × 3.0 mm. The experiment was split into four runs, one for each condition (randomized order), producing 1376 volumes in total. A WM run lasted 750 sec and included acquisition of 379 volumes. An ATT run lasted 618 sec and included 309 volumes. A T1-weighted 3-D magnetization prepared rapid gradient echo data set (256-mm FOV, 265 × 256 grid) was acquired in the same session as the functional images.

Data Analysis

Behavioral Data

Behavioral data (accuracy and RTs) were analyzed using SPSS. For assessment of the baseline data, repeated-measures ANOVAs with Task (WM, ATT), Set size (4, 5, 6), and Distractor (absent, present) as within-subject factors and Group (FT, MT) as between-subject factor were performed. To assess training effects, the additional factor Time (pre, post) was included.

Functional Data

The fMRI data were analyzed with SPM8 (Wellcome Department of Cognitive Neurology, London, UK; www.fil.ion.ucl.ac.uk/spm/software/spm8/). Preprocessing included slice-time correction, realignment of functional images, coregistration between anatomical and functional scans, normalization of all scans to the Montreal Neurological Institute coordinates (interpolating to 3-mm cubic voxels), segmentation, and spatial smoothing with a 6-mm Gaussian kernel. Single-participant analyses were performed on the first level, where the pre–post contrasts between conditions were fixed. In the second-level analysis, FT and MT groups were contrasted. A general linear model was estimated by using regressors for each condition, set size, and blocks separately (ATT+, ATT−, WM+, WM−; four, five, and six targets; first or second block), resulting in 24 regressors in total. Regressors were convolved with a canonical hemodynamic response function and its derivative. Only correct trials were used for general linear model estimation. Percent correct and standard deviations for MT participants were as follows: WM+ pre = 69.29 (11.59) and WM+ post = 75.99 (13.85); for FT, WM+ pre = 68.06 (12.31) and WM+ post = 75.83 (10.36). Three fMRI data sets were discarded because of claustrophobia (participant left scanner during the experiment) and technical problems with the scanner.

To assess the training effects, presession and postsession data of WM+ trials were contrasted (post > pre for increase and pre > post for decrease over training period) separately for both training groups. All p values were set to p < .001, uncorrected.

RESULTS

Behavioral Data at Baseline

Set size and the presence of distractors had a significant effect both on accuracy and RTs (Table 2): the more items had to be processed, the lower accuracy rates (F(2, 56) = 48.21, p = .000) and the longer RTs (F(2, 56) = 60.62, p = .000) became in both ATT and WM trials. The addition of distractors had similar effects and reduced accuracy (F(1, 28) = 64.29, p = .000) and prolonged RTs (F(1, 28) = 82.65, p = .000). Accuracy was lower in ATT than in WM trials (mean ± SD: ATT, 64.4 ± 11%; WM, 73.6 ± 10%; paired t test, t(1, 28) = −6.03, p = .000). At baseline, there was no difference between groups neither for accuracy (F(1, 28) = 0.31, p = .861) nor for RT data (F(1, 28) = 0.87, p = .363).

Table 2. 

Influence of Display Size with (+) and without (−) Distractors on Accuracy in Attention (ATT) and WM Task

Four TargetsFive TargetsSix Targets
ATT− 69.7 (1698) 67.2 (1831) 63.8 (1995) 
ATT+ 66.1 (1835) 61.5 (2050) 58.3 (2144) 
WM− 84.3 (972) 82.5 (978) 69.0 (984) 
WM+ 72.3 (1068) 69.7 (1047) 64.0 (1035) 
Four TargetsFive TargetsSix Targets
ATT− 69.7 (1698) 67.2 (1831) 63.8 (1995) 
ATT+ 66.1 (1835) 61.5 (2050) 58.3 (2144) 
WM− 84.3 (972) 82.5 (978) 69.0 (984) 
WM+ 72.3 (1068) 69.7 (1047) 64.0 (1035) 

Performance: percent correct (RT in milliseconds). For + conditions, the same number of distractors was added.

Training Effects

Both interventions improved general task performance (F(1, 28) = 51.72, p = .000), and the trainings abolished the difference between ATT and WM accuracy rates (Trial type × Time interaction with F(1, 28) = 13.76, p = .001) so that trained participants reached the same level in ATT as in WM after training (post-ATT: 77.4 ± 2.0%; post-WM: 78.9 ± 2.0%; t(1, 28) = −0.66, p = .51). In Figure 1, the training effects are presented separately for ATT and WM trials.

In ATT trials (ATT+ and ATT−), both groups showed a training effect (F(1, 28) = 57.18, p = .000; pre–post effects for FT group: t(1, 13) = −7.45, p = .000; for MT group: t(1, 13) = −3.51, p = .004). The Group × Time interaction indicates that the increase in accuracy was significantly higher for the FT group than for the MT group (F(1, 28) = 4.98, p = .034). Although there was no difference in the premeasurement, we observed a 16.9% higher accuracy for the FT group than for the MT group after training (t(1, 27) = −1.96, p = .030). The Group × Time × Distractor interaction failed to reach significance demonstrating that the trainings affected distractor-present and distractor-absent trials to the same extent.

In WM trials, there was also a main effect of Time (F(1, 28) = 12.14, p = .002) that did not differ between groups (Time × Group with F(1, 28) = 0.1, p = .756; pre–post effects for the FT group: t(1, 13) = −2.303, p = .037; for the MT group: t(1, 13) = −2.95, p = .011). No differences were observed between trials with and without distractors for both groups (all ps > .05). As WM+ trials were of special interest for the fMRI analysis outlined below, these trials were submitted to a separate analysis, which revealed a main effect of Time (F(1, 28) = 15.36, p = .001) but no Time × Group interaction (F(1, 28) = 0.084, p = .774). Instead, both groups improved their performance from pre to post in the critical WM+ condition to the same extent (FT: t(1, 13) = −2.73, p = .016; MT: t(1, 13) = −2.91, p = .012).

Filter and Storage Activity

We examined the brain activity changes from pre to post in the two training groups during the performance of the WM+ trials, which included both the necessity to filter and store information in WM (Figure 2). For the FT group, an increase in activity was observed in the right middle frontal gyrus (MFG; maximum at 33, −1, 46; p = .000) and in the cuneus within the occipital cortex (OCC; maximum at 3, −82, 16; p = .000). There was no training-associated activity decrease in the FT group. The MT group (Figure 1) displayed an increase in bilateral PPC (maxima at 30, −64, 52, p = .000, and −30, −61, 55, p = .001). An activity decrease was observed in the left and right BG (maxima at 15, 11, −5, p < .05 FWE; 33, 17, −17, p = .000, and at −18, 11, 10, p = .000, and −21, 2, −11, p = .000).

Figure 2. 

Pre- and post-fMRI contrasts for the WM+ trials for the FT group (A) and the MT group (C). Activation increases (post > pre) are depicted in reddish colors; decreases (pre > post) are in bluish colors (p < .001). Increase in activity was observed for the FT group (A) in the right MFG and the OCC; for the MT group (C), in bilateral PPC. MT led also to an activity decrease in the left and right BG. For the named regions, activation changes from premeasurement to postmeasurement expressed in beta values are shown in B and D. BG: here, putamen and globus pallidus; OCC: here, cuneus.

Figure 2. 

Pre- and post-fMRI contrasts for the WM+ trials for the FT group (A) and the MT group (C). Activation increases (post > pre) are depicted in reddish colors; decreases (pre > post) are in bluish colors (p < .001). Increase in activity was observed for the FT group (A) in the right MFG and the OCC; for the MT group (C), in bilateral PPC. MT led also to an activity decrease in the left and right BG. For the named regions, activation changes from premeasurement to postmeasurement expressed in beta values are shown in B and D. BG: here, putamen and globus pallidus; OCC: here, cuneus.

Next, we assessed whether training-associated changes in neuronal activity were related to changes in behavioral performance. To do so, we defined the abovementioned brain areas as ROIs and correlated the training-induced activity changes in all of these ROIs with the change in hit rates in the WM tasks. Only in one brain region were the beta values correlated with hit rates, and this was the MFG (r = .429, p = .032).

DISCUSSION

This study investigated the effects of two different cognitive training regimens1 on WM performance and related brain activity. Both training protocols, FT and MT, were able to improve performance in the trained functions (near transfer). However, the improvement in accuracy in the attention task was larger after FT than after MT. In contrast, both trainings improved performance in the memory tasks to the same extent, although only one group had received dedicated MT. In other words, the far transfer effect on the untrained function was larger after FT than after MT.2 The improvements were observed for WM+ trials in which both storage and filtering were required, but also for WM− trials, which did not require filtering of distractors. Hence, our findings do not support the simple assumption that FT influences only trials with distractors and MT has impact only on trials in which information storage is necessary. Instead, our results suggest that more complex near and far transfer effects from the trained to the untrained function were taking place. FT, for example, seemed to promote general selection abilities, which then drove memory encoding efficiency. In this respect, it is noteworthy that, also in trials without distractors, it was necessary to filter out unnecessary sensory input such as environmental noise, visual information from beyond the computer screen, distracting thoughts, and so on.

After we had established that both trainings were able to improve WM, we asked whether these behavioral changes were conveyed by the same neuronal mechanisms or whether the trainings interfered with different neuronal networks. To answer this question, fMRI data before and after training were analyzed. The results point to the latter notion as training-related activity changes occurred in different networks and in different directions for both groups suggesting that different neuronal processes underlay the similar behavioral effects. This interpretation, however, needs to be made with care as the between-group contrasts had failed to reach significance. FT induced an activity increase in prefrontal (MFG) and occipital regions. The MFG has been proposed as a key region of the neuronal gatekeeper network, which hinders irrelevant information from entering WM (McNab & Klingberg, 2008). pFC has also been suggested to top–down control activity in visual areas in OCC, thereby optimizing visual perception (Hopfinger, Buonocore, & Mangun, 2000; Ress et al., 2000). Hence, the FT-related activity increase in both frontal cortex and OCC is likely to reflect a strengthened neuronal loop for the effective control of visual information processing. The importance of the MFG for effective WM is underlined further by the observed correlation between training-induced activity changes in this region and associated improvement in behavioral hit rates. FT did not induce activity decreases anywhere in the brain. It could have been assumed that improved filtering should have reduced neuronal activity in visual areas where less processing resources needed to be devoted to distractors. However, this might have affected small retinotopic visual subregions, which we were unable to detect given that no retinotopic mapping was performed. Even then, suppressed activity had not necessarily to be expected: Filtering could also affect later processing steps or include an active tagging process in the visual system, and the observed activity increases in the OCC might indeed point to the latter. MT, on the other hand, led to a pronounced activity decrease in the BG and also led to an activity increase in the PPC. The PPC has been suggested as a memory storage node, as its activity levels correlate with the number of items one holds in memory (Vogel & Machizawa, 2004).

The PPC has also been shown to be activated when participants unnecessarily store irrelevant information in memory (Kelley & Yantis, 2010; McNab & Klingberg, 2008; Vogel et al., 2005). The BG constitute the subcortical part of the neuronal gatekeeper network. Hence, an inverse relationship between activity levels in the BG and the PPC in low-capacity participants has been interpreted as a failure of the BG to effectively prevent unnecessary storage of information in the PPC (McNab & Klingberg, 2008). Here, we observed that MT shifted neuronal activity from the filtering (BG) toward the storage (PPC) network. This may indicate that MT simply stimulated storage of any kind of information. Alternatively, activity reductions in the BG may be related to learning (e.g., Delgado et al., 2005; Seger & Cincotta, 2005) as the BG are known to be involved in stimulus–response and probabilistic category learning tasks in humans (Packard & Knowlton, 2002; Jog, Kubota, Connolly, Hillegaart, & Graybiel, 1999). However, whereas we observed activity changes in the putamen and pallidum part of the BG, the learning effects in the earlier studies were found mainly in the caudate nucleus. Furthermore, they were suggested to be mediated by pFC. In our study, however, MT did not drive activity in pFC. Hence, we favor the idea that the observed decrease in BG activity together with the increase in PPC after MT reflects a strategic change in stimulus encoding, namely, a bias away from filtering and toward storage. FT, on the other hand, by strengthening the fronto-occipital processing loop likely improved WM by hindering irrelevant information from entering memory. From an ecological point of view, this seems to be the more efficient way for enhancing memory capacity.3

A couple of possible limitations of this study need to be accounted for. First, we did not include a passive control group, and there was no control for unspecific repetition effects. However, a passive control group would still leave it unclear whether training effects in the active groups were related to the specific training regimens or simply to the fact that our participants were working with a computer and had social interaction with the supervisor on 5 consecutive days. Instead, we tried to keep the exposure effects caused by the stimulus material (handling with colored bars on different displays, responding by button press) similar between groups to extract attention and memory-specific training effects. Although, with this design, we cannot decompose specific training effects from exposure or repetition effects within a group, the fact that both trainings induced different behavioral and neurophysiological changes across groups excludes the possibility that all training effects were simply based on repetition. Second, although we tried to keep the training protocols comparable regarding basic perceptual requirements, this goal could not be fully accomplished. For example, the number of stimuli to be processed at a given time was twice as high in the attention task with its double displays. This could explain the observed baseline performance differences between ATT and WM trials and could also have made FT more effortful. Furthermore, FT with its ever changing cues involved frequent task switches, whereas this was not necessary in MT where cues were always black squares. On the other hand, effortful processing and task switching can be considered hallmarks of attentional selection (Jost, De Baene, Koch, & Brass, 2013) and therefore do not necessarily comprise the obtained results and their interpretation. What exact subcomponent of our FT protocol induced the observed effects, however, must remain a question for future research.

More research should focus on the role of selective attention as a means to improve memory. In the past, most MT studies have revealed rather sobering results, especially in terms of inducing transfer effects to untrained cognitive functions (Sprenger et al., 2013; Owen et al., 2010; Holmes, Gathercole, & Dunning, 2009; Dahlin, Neely, et al., 2008; Dahlin, Nyberg, et al., 2008; Li et al., 2008). MT is also not recommended for older adults or patients experiencing memory deficits resulting from neurodegenerative disease such as Alzheimer's (Bahar-Fuchs, Clare, & Woods, 2013; Sitzer, Twamley, & Jeste, 2006). For these patients, MT is generally frustrating as it confronts them continuously with their deficits. Here, training of a function that is less impaired but has the potential to improve memory nevertheless may be more encouraging and motivating for the patients and their caregivers.

In summary, our results support the initial hypothesis that FT does not only improve the handling of dedicated distractor stimuli but selection in general and—as a logical consequence—memory encoding processes.

Acknowledgments

This work was funded by DFG grant Mu1364/4 to N. M.

Reprint requests should be sent to Notger G. Müller, Deutsches Zentrum für Neurodegenerative Erkrankungen (DZNE) Magdeburg, Leipziger Straße 44, Haus 64, 39120 Magdeburg, Germany, or via e-mail: Notger.Mueller@dzne.de.

Notes

1. 

Although our intervention certainly involved practice (i.e., getting acquainted with the task procedures), it involved aspects that went beyond a mere practice and qualify the program as training: The exercised tasks were different from the pre/post assessment tasks, and they aimed at improving specific subcomponents of WM rather than simply boosting task performance per se.

2. 

Note that far transfer here is based on the definition by Noack, Lövdén, Schmiedek, and Lindenberger (2009), who postulated that fat transfer would relate to effects on other, untrained cognitive constructs.

3. 

Note that, although our trainings were intended to change the way/strategy participants relied on filtering or storage during encoding, we at the same time had taken special care that our training regimen did not foster more complex encoding strategies such as chunking, forming semantic associations by grouping of stimuli to complex objects, or simply counting stimuli to perform better.

REFERENCES

Bahar-Fuchs
,
A.
,
Clare
,
L.
, &
Woods
,
B.
(
2013
).
Cognitive training and cognitive rehabilitation for mild to moderate Alzheimer's disease and vascular dementia
.
Cochrane Database System Review 6: CD003260
.
DOI:10.1002/14651858.CD003260.pub2.
Cowan
,
N.
(
2000
).
The magical number 4 in short-term memory: A reconsideration of mental storage capacity
.
Behavioral and Brain Sciences
,
24
,
87
185
.
Cowan
,
N.
, &
Morey
,
C. C.
(
2009
).
NIH public access
.
Brain
,
10
,
139
141
.
Dahlin
,
E.
,
Neely
,
A. S.
,
Larsson
,
A.
,
Bäckman
,
L.
, &
Nyberg
,
L.
(
2008
).
Transfer of learning after updating training mediated by the striatum
.
Science
,
320
,
1510
1512
.
Dahlin
,
E.
,
Nyberg
,
L.
,
Bäckman
,
L.
, &
Neely
,
A. S.
(
2008
).
Plasticity of executive functioning in young and older adults: Immediate training gains, transfer, and long-term maintenance
.
Psychology and Aging
,
23
,
720
730
.
Delgado
,
M. R.
,
Miller
,
M. M.
,
Inati
,
S.
, &
Phelps
,
E. A.
(
2005
).
An fMRI study of reward-related probability learning
.
Neuroimage
,
24
,
862
873
.
Hempel
,
A.
,
Giesel
,
F. L.
,
Garcia Caraballo
,
N. M.
,
Amann
,
M.
,
Meyer
,
H.
,
Wustenberg
,
T.
, et al
(
2004
).
Plasticity of cortical activation related to working memory during training
.
American Journal of Psychiatry
,
161
,
745
747
.
Holmes
,
J.
,
Gathercole
,
S. E.
, &
Dunning
,
D. L.
(
2009
).
Adaptive training leads to sustained enhancement of poor working memory in children
.
Developmental Science
,
12
,
F9
F15
.
Hopfinger
,
J. B.
,
Buonocore
,
M. H.
, &
Mangun
,
G. R.
(
2000
).
The neural mechanisms of top–down attentional control
,
3
,
284
291
.
Jog
,
M. S.
,
Kubota
,
Y.
,
Connolly
,
C. I.
,
Hillegaart
,
V.
, &
Graybiel
,
A. M.
(
1999
).
Building neural representations of habits
.
Science
,
286
,
1745
1749
.
Jost
,
K.
,
De Baene
,
W.
,
Koch
,
I.
, &
Brass
,
M.
(
2013
).
A review of the role of cue processing in task switching
.
Zeitschrift für Psychologie
,
221
,
5
14
.
Kelley
,
T. A.
, &
Yantis
,
S.
(
2010
).
Neural correlates of learning to attend
.
Frontiers in Human Neuroscience
,
4
,
216
.
Li
,
S. C.
,
Schmiedek
,
F.
,
Huxhold
,
O.
,
Röcke
,
C.
,
Smith
,
J.
, &
Lindenberger
,
U.
(
2008
).
Working memory plasticity in old age: Practice gain, transfer, and maintenance
.
Psychology and Aging
,
23
,
731
742
.
Luck
,
S. J.
, &
Vogel
,
E. K.
(
2013
).
Visual working memory capacity: From psychophysics and neurobiology to individual differences
.
Trends in Cognitive Sciences
,
17
,
391
400
.
McNab
,
F.
, &
Klingberg
,
T.
(
2008
).
Prefrontal cortex and basal ganglia control access to working memory
.
Nature Neuroscience
,
11
,
103
107
.
Noack
,
H.
,
Lövdén
,
M.
,
Schmiedek
,
F.
, &
Lindenberger
,
U.
(
2009
).
Cognitive plasticity in adulthood and old age: Gauging the generality of cognitive intervention effects
.
Restorative Neurology and Neuroscience
,
27
,
435
453
.
Owen
,
A. M.
,
Hampshire
,
A.
,
Grahn
,
J. A.
,
Stenton
,
R.
,
Dajani
,
S.
,
Burns
,
A. S.
, et al
(
2010
).
Putting brain training to the test
.
Nature
,
465
,
775
778
.
Packard
,
G. P.
, &
Knowlton
,
B. J.
(
2002
).
Learning and memory functions of the basal ganglia
.
Annual Review of Neuroscience
,
25
,
563
593
.
Ress
,
D.
,
Backus
,
B. T.
, &
Heeger
,
D. J.
(
2000
).
Activity in primary visual cortex predicts performance in a visual detection task
.
Nature Neuroscience
,
3
,
940
945
.
Seger
,
C. A.
, &
Cincotta
,
C. M.
(
2005
).
The roles of the caudate nucleus in human classification learning
.
Journal of Neuroscience
,
25
,
2941
2951
.
Sitzer
,
D. I.
,
Twamley
,
E. W.
, &
Jeste
,
D. V.
(
2006
).
Cognitive training in Alzheimer's disease: A meta-analysis of the literature
.
Acta Psychiatrica Scandinavica
,
114
,
75
90
.
Sprenger
,
A. M.
,
Atkins
,
S. M.
,
Bolger
,
D. J.
,
Harbison
,
J. I.
,
Novick
,
J. M.
,
Chrabaszcz
,
J. S.
, et al
(
2013
).
Training working memory: Limits of transfer
.
Intelligence
,
41
,
638
663
.
Todd
,
J. J. A. Y.
, &
Marois
,
R.
(
2005
).
Posterior parietal cortex activity predicts individual differences in visual short-term memory capacity
.
Vision Research
,
5
,
144
155
.
Vogel
,
E. K.
, &
Machizawa
,
M. G.
(
2004
).
Neural activity predicts individual differences in visual working memory capacity
.
Nature
,
428
,
748
751
.
Vogel
,
E. K.
,
McCollough
,
A. W.
, &
Machizawa
,
M. G.
(
2005
).
Neural measures reveal individual differences in controlling access to working memory
.
Nature
,
438
,
500
503
.