Attention induces surround suppression in visual working memory

Serial dependence: A matter of memory load

In serial dependence, perceptual decisions are biased towards stimuli encountered in the recent past. Here, we investigate whether and how serial dependence is affected by the availability of visual working memory (VWM) resources. In two experiments, participants reproduced the orientation of a series of stimuli. On alternating trials, we included an additional VWM task with randomly varying levels of load. Serial dependence was not only affected by the additional load task but also clearly modulated by the level of load: a high load in the previous trial reduced serial dependence while a high load in the present increased it. These results were independent of the effects of VWM load on the precision of reproduction responses. Our findings provide new insights into the mechanisms that may regulate serial dependence, revealing its intimate link with VWM resources.

Significance statement

Our perception, thoughts, and behavior are continuously influenced by recent events. For instance, the way we process and understand current visual information depends on what we have seen in the preceding seconds, a phenomenon known as serial dependence. The precise mechanisms and factors involved in serial dependence are still unclear. Here, we demonstrated that working memory resources are a crucial component. Specifically, when we are currently experiencing a heavy memory load, the influence of prior stimuli becomes stronger. Conversely, when prior stimuli were shown under a high memory load, their influence was reduced. These findings highlight the importance of working memory resources in shaping our interpretation of the present based on the recent past.

Spatial Configuration Effects on the Dissociation between Active and Latent States in Visual Working Memory

Visual working memory includes both active, recallable items and latent items not directly available for recall. During the online manipulation of active working memory, latent items gain robust retention. According to the dissociation account, active and passive memories exist in independent states, both of which take up their own cognitive resources. However, it is unclear whether dissociation is a universal phenomenon during memory maintenance. Given that memory information is retained as a binding of identity and location, the spatial proximity of memory items might shield the dissociation property. To test this, we adopted a retro-cue memory task where the cued and uncued items were separated in the active and latent states. In Experiment 1, the memory items were presented at a relatively large spatial distance. The results supported the dissociation account for well-separated items. However, Experiment 2 rejected the dissociation for closer-in items, possibly because items in visual working memory were spatially labeled. These findings suggest that while memory maintenance generally conforms to the dissociation account, the spatial configuration of memory items can affect the dissociation property between the active and latent neural states in visual working memory.

Object-based selection in visual working memory

Attentional mechanisms in perception can operate over locations, features, or objects. However, people direct attention not only towards information in the external world, but also to information maintained in working memory. To what extent do perception and memory draw on similar selection properties? Here we examined whether principles of object-based attention can also hold true in visual working memory. Experiment 1 examined whether object structure guides selection independently of spatial distance. In a memory updating task, participants encoded two rectangular bars with colored ends before updating two colors during maintenance. Memory updates were faster for two equidistant colors on the same object than on different objects. Experiment 2 examined whether selection of a single object feature spreads to other features within the same object. Participants memorized two sequentially presented Gabors, and a retro-cue indicated which object and feature dimension (color or orientation) would be most relevant to the memory test. We found stronger effects of object selection than feature selection: accuracy was higher for the uncued feature in the same object than the cued feature in the other object. Together these findings demonstrate effects of object-based attention on visual working memory, at least when object-based representations are encouraged, and suggest shared attentional mechanisms across perception and memory.

The extent of center-surround inhibition for colored items in working memory

Remembering a color suppresses the representations of similar colors, but not of distinct colors, producing a center-surround inhibition (CSI) to resolve the competition between similar colors. In this study, three probe experiments were conducted to investigate the extent of CSI for colored items in working memory (WM). In Experiments 1 and 2, two WM items (distance 0°, 20°, 40°, or 60° in color space) were presented sequentially, one of which was cued to compare with the probe (matched or non-matched). The probe distance between the non-matched probe (NP) and cued WM item was 30° in Experiment 1 and 30°, 60°, or 90° in Experiment 2. Results for matched probe (MP) revealed that two WM items might produce a maximal CSI at distance 20°, and fall outside each other's inhibitory surround at distance 40°. However, the CSI was not found in the NP conditions (i.e., distance 30°, 60°, or 90°) in both Experiments 1 and 2, suggesting that the NP might be unsuitable for investigating the CSI in WM. In Experiment 3, participants were asked to discriminate which WM item was matched with the probe (no NP conditions). RTs were slowest at distance 20°, but were almost equal across distance 30°, 40°, 50°, or 60°. These results demonstrated that two WM items might produce a maximal CSI at distance 20°, and begin to fall outside each other's inhibitory surround at distance 30°.

Reevaluating the sensory recruitment model by manipulating crowding in visual working memory representations

The prominent sensory recruitment model argues that visual working memory (WM) is maintained via representations in the same early visual cortex brain regions that initially encode sensory stimuli, either in the identical neural populations as perceptual representations or in distinct neural populations. While recent research seems to reject the former (strong) sensory recruitment model, the latter (flexible) account remains plausible. Moreover, this flexibility could explain a recent result of high theoretical impact (Harrison & Bays, The Journal of Neuroscience, 38 (12), 3116-3123, 2018) - a failure to observe interactions between items held in visual WM - that has been taken to reject the sensory recruitment model. Harrison and Bays (The Journal of Neuroscience, 38 (12), 3116-3123, 2018) tested the sensory recruitment model by comparing the precision of memoranda in radially and tangentially oriented memory arrays. Because perceptual visual crowding effects are greater in radial than tangential arrays, they reasoned that a failure to observe such anisotropy in WM would reject the sensory recruitment model. In the present Registered Report or Replication, we replicated their study with greater sensitivity and extended their task by controlling a potential strategic confound. Specifically, participants might remap memory items to new locations, reducing interactions between proximal memoranda. To combat remapping, we cued participants to report either a memory item or its precise location - with this report cue presented only after a memory maintenance period. Our results suggest that, similar to visual perceptual crowding, location-bound visual memoranda interact with one another when remapping is prevented. Thus, our results support at least a flexible form of the sensory recruitment model.

The profile of attentional modulation to visual features

Although it is well established that feature-based attention (FBA) can enhance an attended feature, how it modulates unattended features remains less clear. Previous studies have generally supported either a graded profile as predicted by the feature-similarity gain model or a nonmonotonic profile predicted by the surround suppression model. To reconcile these different views, we systematically measured the attentional profile in three basic feature dimensions-orientation, motion direction, and spatial frequency. In three experiments, we instructed participants to detect a coherent feature signal against noise under attentional or neutral condition. Our results support a nonmonotonic hybrid model of attentional modulation consisting of feature-similarity gain and surround suppression for orientation and motion direction. For spatial frequency, we also found a similar nonmonotonic profile for higher frequencies than the attended frequency, but a lack of attentional modulation for lower frequencies than the attended frequency. The current findings can reconcile the discrepancies in the literature and suggest the hybrid model as a new framework for attentional modulation in feature space. In addition, a computational model incorporating known properties of spatial frequency channels and attentional modulations at the neural level reproduced the asymmetric attentional modulation, thus revealing a connection between surround suppression and the basic neural architecture of an early visual system.