Stochastic sampling provides a unifying account of visual working memory limits

Comparing retro-cue benefit mechanisms in visual working memory: completely valid vs. highly valid retro-cues

Visual working memory (VWM) plays a crucial role in temporarily maintaining and manipulating visual information. Retro-cue benefit (RCB) refers to the enhancement of memory performance when attention is directed toward a subset of items in VWM after their initial encoding. Our recent electroencephalogram (EEG) studies indicate that cue validity affects the mechanisms underlying RCB formation. However, previous research has not thoroughly examined whether these mechanisms differ between completely valid and highly valid cue conditions. This study investigates the consistency of RCB mechanisms under conditions of complete (100%) and high (80%) retro-cue validity. We manipulated retro-cue validity and examined cognitive processing mechanisms under different validity conditions using EEG. Specifically, we focused on the N2pc component, which reflects attentional resource allocation, and the contralateral delay activity (CDA) component, which reflects the quantity of information retained in VWM. The results, encompassing both behavioral and event-related potential (ERP) findings, show that participants in both the 100% and 80% cue validity conditions exhibit robust RCB. Notably, the degree of RCB remains consistent across these conditions, indicating that participants utilize retro-cues to enhance VWM performance to the same extent. In the 80% cue validity condition, a significant retro-cue cost (RCC) was observed, indicating that participants selectively discarded uncued items from VWM. In invalid trials, response accuracy drops to chance levels, supporting the removal hypothesis. ERP results reveal that attentional resource allocation (N2pc) and the quantity of retained information (CDA) remain uniform across cue validity conditions. The mechanism responsible for RCB formation appears to involve an all-or-nothing process of discarding uncued information rather than a flexible resource allocation strategy. This study provides insights into attention allocation and information-processing mechanisms in VWM, suggesting that conclusions drawn from tasks with completely valid retro-cues can be integrated with findings from highly valid cue tasks. These findings also illuminate the flexibility of internal attentional resource allocation during RCB formation and contribute to our understanding of attention processes in VWM.

Is working memory domain-general or domain-specific?

Given the fundamental role of working memory (WM) in all domains of cognition, a central question has been whether WM is domain-general. However, the term 'domain-general' has been used in different, and sometimes misleading, ways. By reviewing recent evidence and biologically plausible models of WM, we show that the level of domain-generality varies substantially between three facets of WM: in terms of computations, WM is largely domain-general. In terms of neural correlates, it contains both domain-general and domain-specific elements. Finally, in terms of application, it is mostly domain-specific. This variance encourages a shift of focus towards uncovering domain-general computational principles and away from domain-general approaches to the analysis of individual differences and WM training, favoring newer perspectives, such as training-as-skill-learning.

Removing information from working memory with a delay: Effective but not beneficial

Ideally, removing outdated information from working memory (WM) should have two consequences: The removed content should be less accessible (removal costs), and other WM content should benefit from the freeing up of WM capacity (removal benefits). Robust removal benefits and removal costs have been demonstrated when people are told to forget items shortly after they were encoded (immediate removal). However, other studies suggest that people might be unable to selectively remove items from an already encoded set of items (delayed removal). In two experiments (n = 219; n = 241), we investigated the effectiveness and consequences of delayed removal by combining a modified version of Ecker's et al. (Journal of Memory and Language, 74, 77-90, 2014) letter updating task with a directed-forgetting in WM paradigm. We found that while delayed removal resulted in reduced memory for the to-be-forgotten item-location relations (removal costs), it failed to enhance performance for existing WM content. This contrasts sharply with immediate removal, where removal benefits can be observed. A fine-grained analysis of removal benefits shows that removal from WM proactively facilitates the subsequent encoding of new information but does not retroactively aid stored WM content.

Visual working memory for natural scenes: challenges and opportunities

Visual working memory is a fundamental cognitive process that people use thousands of times each day as they engage in visually guided behavior. Thus, it is important to understand how the natural visual input-which consists of complex, spatially organized, continuously varying features-is represented in visual working memory. However, most research has used arrays of discrete, artificial objects defined by simple features, and existing formal models of visual working memory cannot be applied to natural scenes. In this paper, we identify 3 key aspects of natural scenes that are not captured by existing formal models of visual working memory, along with 2 distinct types of attention that must be considered. The goal is to clearly define the challenges and opportunities for moving models of visual working memory from arrays of artificial objects to natural scenes.

Visual working memory models of delayed estimation do not generalize to whole-report tasks

Whole-report working memory tasks provide a measure of recall for all stimuli in a trial and afford single-trial analyses that are not possible with single-report delayed estimation tasks. However, most whole-report studies assume that trial stimuli are encoded and reported independently, and they do not consider the relationships between stimuli presented and reported within the same trial. Here, we present the results of two independently conducted whole-report experiments. The first dataset was recorded by Adam, Vogel, and Awh (2017) and required participants to report color and orientation stimuli using a continuous response wheel. We recorded the second dataset, which required participants to report color stimuli using a set of discrete buttons. We found that participants often group their reports by color similarity, contradicting the assumption of independence implicit in most encoding models of working memory. Next, we showed that this behavior was consistent across participants and experiments when reporting color but not orientation, two circular variables often assumed to be equivalent.Finally, we implemented an alternative to independent encoding where stimuli are encoded as a hierarchical Bayesian ensemble and found that this model predicts biases that are not present in either dataset. Our results suggest that assumptions made by both independent and hierarchical ensemble encoding models-which were developed in the context of single-report delayed estimation tasks-do not hold for the whole-report task. This failure to generalize highlights the need to consider variations in task structure when inferring fundamental principles of visual working memory.

Representation and computation in visual working memory

The ability to sustain internal representations of the sensory environment beyond immediate perception is a fundamental requirement of cognitive processing. In recent years, debates regarding the capacity and fidelity of the working memory (WM) system have advanced our understanding of the nature of these representations. In particular, there is growing recognition that WM representations are not merely imperfect copies of a perceived object or event. New experimental tools have revealed that observers possess richer information about the uncertainty in their memories and take advantage of environmental regularities to use limited memory resources optimally. Meanwhile, computational models of visuospatial WM formulated at different levels of implementation have converged on common principles relating capacity to variability and uncertainty. Here we review recent research on human WM from a computational perspective, including the neural mechanisms that support it.

The Determining Role of Covariances in Large Networks of Stochastic Neurons

Biological neural networks are notoriously hard to model due to their stochastic behavior and high dimensionality. We tackle this problem by constructing a dynamical model of both the expectations and covariances of the fractions of active and refractory neurons in the network's populations. We do so by describing the evolution of the states of individual neurons with a continuous-time Markov chain, from which we formally derive a low-dimensional dynamical system. This is done by solving a moment closure problem in a way that is compatible with the nonlinearity and boundedness of the activation function. Our dynamical system captures the behavior of the high-dimensional stochastic model even in cases where the mean-field approximation fails to do so. Taking into account the second-order moments modifies the solutions that would be obtained with the mean-field approximation and can lead to the appearance or disappearance of fixed points and limit cycles. We moreover perform numerical experiments where the mean-field approximation leads to periodically oscillating solutions, while the solutions of the second-order model can be interpreted as an average taken over many realizations of the stochastic model. Altogether, our results highlight the importance of including higher moments when studying stochastic networks and deepen our understanding of correlated neuronal activity.

A Neural Decision Signal during Internal Sampling from Working Memory in Humans

How humans transform sensory information into decisions that steer purposeful behavior is a central question in psychology and neuroscience that is traditionally investigated during the sampling of external environmental signals. The decision-making framework of gradual information sampling toward a decision has also been proposed to apply when sampling internal sensory evidence from working memory. However, neural evidence for this proposal remains scarce. Here we show (using scalp EEG in male and female human volunteers) that sampling internal visual representations from working memory elicits a scalp EEG potential associated with gradual evidence accumulation-the central parietal positivity. Consistent with an evolving decision process, we show how this signal (1) scales with the time participants require to reach a decision about the cued memory content and (2) is amplified when having to decide among multiple contents in working memory. These results bring the electrophysiology of decision-making into the domain of working memory and suggest that variability in memory-guided behavior may be driven (at least in part) by variations in the sampling of our inner mental contents.

A dynamic neural resource model bridges sensory and working memory

Probing memory of a complex visual image within a few hundred milliseconds after its disappearance reveals significantly greater fidelity of recall than if the probe is delayed by as little as a second. Classically interpreted, the former taps into a detailed but rapidly decaying visual sensory or 'iconic' memory (IM), while the latter relies on capacity-limited but comparatively stable visual working memory (VWM). While iconic decay and VWM capacity have been extensively studied independently, currently no single framework quantitatively accounts for the dynamics of memory fidelity over these time scales. Here, we extend a stationary neural population model of VWM with a temporal dimension, incorporating rapid sensory-driven accumulation of activity encoding each visual feature in memory, and a slower accumulation of internal error that causes memorized features to randomly drift over time. Instead of facilitating read-out from an independent sensory store, an early cue benefits recall by lifting the effective limit on VWM signal strength imposed when multiple items compete for representation, allowing memory for the cued item to be supplemented with information from the decaying sensory trace. Empirical measurements of human recall dynamics validate these predictions while excluding alternative model architectures. A key conclusion is that differences in capacity classically thought to distinguish IM and VWM are in fact contingent upon a single resource-limited WM store.

Perceptual similarity judgments do not predict the distribution of errors in working memory

Population coding models provide a quantitative account of visual working memory (VWM) retrieval errors with a plausible link to the response characteristics of sensory neurons. Recent work has provided an important new perspective linking population coding to variables of signal detection, including d-prime, and put forward a new hypothesis: that the distribution of recall errors on, for example, a color wheel, is a consequence of the psychological similarity between points in that stimulus space, such that the exponential-like psychophysical distance scaling function can fulfil the role of population tuning and obviate the need to fit a tuning width parameter to recall data. Using four different visual feature spaces, we measured psychophysical similarity and memory errors in the same participants. Our results revealed strong evidence for a common source of variability affecting similarity judgments and recall estimates but did not support any consistent relationship between psychophysical similarity functions and VWM errors. At the group level, the responsiveness functions obtained from the psychophysical similarity task diverged strongly from those that provided the best fit to working memory errors. At the individual level, we found convincing evidence against an association between observed and best-fitting similarity functions. Finally, our results show that the newly proposed exponential-like responsiveness function has in general no advantage over the canonical von Mises (circular normal) function assumed by previous population coding models. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Mapping visual working memory models to a theoretical framework

The body of research on visual working memory (VWM)-the system often described as a limited memory store of visual information in service of ongoing tasks-is growing rapidly. The discovery of numerous related phenomena, and the many subtly different definitions of working memory, signify a challenge to maintain a coherent theoretical framework to discuss concepts, compare models and design studies. A lack of robust theory development has been a noteworthy concern in the psychological sciences, thought to be a precursor to the reproducibility crisis (Oberauer & Lewandowsky, Psychonomic Bulletin & Review, 26, 1596-1618, 2019). I review the theoretical landscape of the VWM field by examining two prominent debates-whether VWM is object-based or feature-based, and whether discrete-slots or variable-precision best describe VWM limits. I share my concerns about the dualistic nature of these debates and the lack of clear model specification that prevents fully determined empirical tests. In hopes of promoting theory development, I provide a working theory map by using the broadly encompassing memory for latent representations model (Hedayati et al., Nature Human Behaviour, 6, 5, 2022) as a scaffold for relevant phenomena and current theories. I illustrate how opposing viewpoints can be brought into accordance, situating leading models of VWM to better identify their differences and improve their comparison. The hope is that the theory map will help VWM researchers get on the same page-clarifying hidden intuitions and aligning varying definitions-and become a useful device for meaningful discussions, development of models, and definitive empirical tests of theories.

The role of recollection and familiarity in visual working memory: A mixture of threshold and signal detection processes

Whether working memory reflects a thresholded recollection process whereby only a limited number of items are maintained in memory, or a signal detection process in which each studied item is increased in familiarity strength, is a topic of considerable debate. A review of visual working memory studies that have examined receiver operating characteristics (ROCs) across a broad set of materials and test conditions indicates that both signal detection and threshold processes contribute to working memory. In addition, the role that these two processes play varies systematically across conditions, such that a threshold process plays a particularly critical role when binary old/new judgments are required, when changes are relatively discrete, and when the hippocampus does not contribute to performance. In contrast, a signal detection process plays a greater role when confidence judgments are required, when the materials or the changes are global in nature, and when the hippocampus contributes to performance. In addition, the ROC results indicate that in standard single-probe tests of working memory, items that are maintained in an active recollected state support both recall-to-accept and recall-to-reject responses; whereas in complex-probe tests, recollection preferentially supports recall-to-reject; and in item-recognition tests it preferentially supports recall-to-accept. Moreover, there is growing evidence that these threshold and strength-based processes are related to distinct states of conscious awareness whereby they support perceiving- and sensing-based responses, respectively. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Revealing visual working memory operations with pupillometry: Encoding, maintenance, and prioritization

Pupillary dynamics reflect effects of distinct and important operations of visual working memory: encoding, maintenance, and prioritization. Here, we review how pupil size predicts memory performance and how it provides novel insights into the mechanisms of each operation. Visual information must first be encoded into working memory with sufficient precision. The depth of this encoding process couples to arousal-linked baseline pupil size as well as a pupil constriction response before and after stimulus onset, respectively. Subsequently, the encoded information is maintained over time to ensure it is not lost. Pupil dilation reflects the effortful maintenance of information, wherein storing more items is accompanied by larger dilations. Lastly, the most task-relevant information is prioritized to guide upcoming behavior, which is reflected in yet another dilatory component. Moreover, activated content in memory can be pupillometrically probed directly by tagging visual information with distinct luminance levels. Through this luminance-tagging mechanism, pupil light responses reveal whether dark or bright items receive more attention during encoding and prioritization. Together, conceptualizing pupil responses as a sum of distinct components over time reveals insights into operations of visual working memory. From this viewpoint, pupillometry is a promising avenue to study the most vital operations through which visual working memory works. This article is categorized under: Psychology > Attention Psychology > Memory Psychology > Theory and Methods.

Exploring retro-cue effects on visual working memory: insights from double-cue paradigm

In the realm of visual working memory research, the retro-cue paradigm helps us study retro-cue effects such as retro-cue benefit (RCB) and retro-cue cost (RCC). RCB reflects better performance with cued items, while RCC indicates poorer performance with uncued items. Despite consistent evidence for RCB, it's still uncertain whether it remains when previously uncued items are cued afterward. Additionally, research findings have been inconsistent. This study combines prior experiments by controlling the proportion of cue types and the number of memory items. Besides, using a CDA index to assess the status of items after the cue appeared. Results showed better performance under the double-cue condition (involving two cues pointing inconsistently with only the second cue being valid) compared to the neutral-cue condition, and better performance under the single-cue condition compared to double-cue. EEG data revealed that after the appearance of the second cue in the double-cue condition, there was a significant increase in CDA wave amplitude compared to the single-cue condition. Behavior results suggests that RCB occurs under double-cue but to a lesser extent than the single-cue. And EEG outcomes indicates that individuals did not remove the uncued item from their visual working memory after the first cue. Instead, they kept it in a passive state and then shifted it to an active state after the appearance of the second cue.

Probabilistic and rich individual working memories revealed by a betting game

When asked to remember a color, do people remember a point estimate (e.g., a particular shade of red), a point estimate plus an uncertainty estimate, or are memory representations rich probabilistic distributions over feature space? We asked participants to report the color of a circle held in working memory. Rather than collecting a single report per trial, we had participants place multiple bets to create trialwise uncertainty distributions. Bet dispersion correlated with performance, indicating that internal uncertainty guided bet placement. While the first bet was on average the most precisely placed, the later bets systematically shifted the distribution closer to the target, resulting in asymmetrical distributions about the first bet. This resulted in memory performance improvements when averaging across bets, and overall suggests that memory representations contain more information than can be conveyed by a single response. The later bets contained target information even when the first response would generally be classified as a guess or report of an incorrect item, suggesting that such failures are not all-or-none. This paradigm provides multiple pieces of evidence that memory representations are rich and probabilistic. Crucially, standard discrete response paradigms underestimate the amount of information in memory representations.

Goal-Dependent Hippocampal Representations Facilitate Self-Control

Hippocampal activity linking past experiences and simulations of the future with current goals can play an important role in decision-making. The representation of information within the hippocampus may be especially critical in situations where one needs to overcome past rewarding experiences and exert self-control. Self-control success or failure may depend on how information is represented in the hippocampus and how effectively the representation process can be modified to achieve a specific goal. We test this hypothesis using representational similarity analyses of human (female/male) neuroimaging data during a dietary self-control task in which individuals must overcome taste temptations to choose healthy foods. We find that self-control is indeed associated with the way individuals represent taste information (valance) in the hippocampus and how taste representations there adapt to align with different goals/contexts. Importantly, individuals who were able to shift their hippocampal representations to a larger degree to align with the current motivation were better able to exert self-control when facing a dietary challenge. These results suggest an alternative or complementary neurobiological pathway leading to self-control success and indicate the need to update the classical view of self-control to continue to advance our understanding of its behavioral and neural underpinnings.SIGNIFICANCE STATEMENT The paper provides a new perspective on what leads to successful self-control at the behavioral and neurobiological levels. Our data suggest that self-control is enhanced when individuals adjust hippocampal processing to align with current goals.

A spatially continuous diffusion model of visual working memory

We present results from five visual working memory (VWM) experiments in which participants were briefly shown between 2 and 6 colored squares. They were then cued to recall the color of one of the squares and they responded by choosing the color on a continuous color wheel. The experiments provided response proportions and response time (RT) measures as a function of angle for the choices. Current VWM models for this task include discrete models that assume an item is either within working memory or not and resource models that assume that memory strength varies as a function of the number of items. Because these models do not include processes that allow them to account for RT data, we implemented them within the spatially continuous diffusion model (SCDM, Ratcliff, 2018) and use the experimental data to evaluate these combined models. In the SCDM, evidence retrieved from memory is represented as a spatially continuous normal distribution and this drives the decision process until a criterion (represented as a 1-D line) is reached, which produces a decision. Noise in the accumulation process is represented by continuous Gaussian process noise over spatial position. The models that fit best from the discrete and resource-based classes converged on a common model that had a guessing component and that allowed the height of the normal memory-strength distribution to vary with number of items. The guessing component was implemented as a regular decision process driven by a flat evidence distribution, a zero-drift process. The combination of choice and RT data allows models that were not identifiable based on choice data alone to be discriminated.

Distinguishing guesses from fuzzy memories: Further evidence for item limits in visual working memory

There is consistent debate over whether capacity in working memory (WM) is subject to an item limit, or whether an unlimited number of items can be held in this online memory system. The item limit hypothesis clearly predicts guessing responses when capacity is exceeded, and proponents of this view have highlighted evidence for guessing in visual working memory tasks. Nevertheless, various models that deny item limits can explain the same empirical patterns by asserting extremely low fidelity representations that cannot be distinguished from guesses. To address this ambiguity, we employed a task for which guess responses elicited a qualitatively distinct pattern from low fidelity memories. Inspired by work from Rouder et al. (2014), we employed an orientation WM task that required subjects to recall the precise orientation of each of six memoranda presented 1 s earlier. The orientation stimuli were created by rotating the position of a "clock hand" inside a circular region that was demarcated by four colored quadrants. Critically, when observers guess with these stimuli, the distribution of responses is biased towards the center of these quadrants, creating a "banded" pattern that cannot be explained by a low precision memory. We confirmed the presence of this guessing pattern using formal model comparisons, and we show that the prevalence of this pattern matches observers' own reports of when they thought they were guessing. Thus, these findings provide further evidence for guessing behaviors predicted by item limit models of WM capacity.

There Is no Theory-Free Measure of "Swaps" in Visual Working Memory Experiments

Visual working memory is highly limited, and its capacity is tied to many indices of cognitive function. For this reason, there is much interest in understanding its architecture and the sources of its limited capacity. As part of this research effort, researchers often attempt to decompose visual working memory errors into different kinds of errors, with different origins. One of the most common kinds of memory error is referred to as a "swap," where people report a value that closely resembles an item that was not probed (e.g., an incorrect, non-target item). This is typically assumed to reflect confusions, like location binding errors, which result in the wrong item being reported. Capturing swap rates reliably and validly is of great importance because it permits researchers to accurately decompose different sources of memory errors and elucidate the processes that give rise to them. Here, we ask whether different visual working memory models yield robust and consistent estimates of swap rates. This is a major gap in the literature because in both empirical and modeling work, researchers measure swaps without motivating their choice of swap model. Therefore, we use extensive parameter recovery simulations with three mainstream swap models to demonstrate how the choice of measurement model can result in very large differences in estimated swap rates. We find that these choices can have major implications for how swap rates are estimated to change across conditions. In particular, each of the three models we consider can lead to differential quantitative and qualitative interpretations of the data. Our work serves as a cautionary note to researchers as well as a guide for model-based measurement of visual working memory processes.

Sustained attention required for effective dimension-based retro-cue benefit in visual working memory

In visual working memory (VWM) tasks, participants' performances can be improved through the use of dimension-based retro-cues, which direct internal attention to prioritize a particular dimension (e.g., color or orientation) of VWM representations even after the stimuli disappear. This phenomenon is known as the dimension-based retro-cue benefit (RCB). The present study investigates whether sustained attention is required for the dimension-based RCB by inserting interference or interruption between the retro-cue and the test array to distract attention. We tested the effects of perceptual interference or cognitive interruption on dimension-based RCB when the interference (Experiments 1 and 2 with masks) or interruption (Experiments 3 and 4 with an odd-even task) occurred concurrently with the stages for the maintenance of prioritized information (long cue-and-interference/interruption interstimulus interval, e.g., Experiments 1 and 3) or the deployment of attention (short cue-and-interference/interruption interstimulus interval, e.g., Experiments 2 and 4). Our results demonstrate that perceptual interference or cognitive interruption attenuates the dimension-based RCB. These findings suggest that sustained attention is necessary for the effective prioritization of a specific dimension of VWM representations.

The spatiotemporal gradient of intrusion errors in continuous outcome source memory: Source retrieval is affected by both guessing and intrusions

Previous research has characterized source retrieval as a thresholded process, which fails on a proportion of trials and leads to guessing, as opposed to a continuous process, in which response precision varies across trials but is never zero. The thresholded view of source retrieval is largely based on the observation of heavy tailed distributions of response errors, thought to reflect a large proportion of "memoryless" trials. In this study, we investigate whether these errors might instead reflect systematic intrusions from other list items which can mimic source guessing. Using the circular diffusion model of decision making, which accounts for both response errors and RTs we found that intrusions account for some, but not all, errors in a continuous-report source memory task. We found that intrusion errors were more likely to come from items studied in nearby locations and times, and were well-described by a spatiotemporal gradient model, but not from semantically or perceptually similar cues. Our findings support a thresholded view of source retrieval but suggest that previous work has overestimated the proportion of guesses which have been conflated with intrusions.

Task Demands Differentially Affect Processing of Intrinsic and Extrinsic Object Features in Working Memory

Some argue that visual working memory operates on integrated object representations. Here, we contend that obligatory feature integration occurs with intrinsic but not extrinsic object features. Working memory for shapes and colors was assessed using a change-detection task with a central test probe, while recording event-related potentials (ERPs). Color was either an intrinsic surface feature of a shape or connected to the shape via a proximal but spatially disjunct extrinsic frame. There were two types of test: The direct test required memory for shape and color; the indirect test required only shape memory. Study-test changes of color were therefore either task-relevant or task-irrelevant. We assessed performance costs and event-related potential (ERP) effects arising from color changes. In the direct test, performance was poorer for extrinsic than intrinsic stimuli; task-relevant color changes elicited enhanced frontal negativity (N2, FN400) for both intrinsic and extrinsic stimuli. In the indirect test, performance costs and ERP effects associated with irrelevant color change were larger for intrinsic than extrinsic stimuli. This suggests intrinsic information is more readily integrated into the working-memory representation and evaluated against the test probe. Findings imply that feature integration is not obligatory under all conditions but influenced by stimulus-driven and task-related focus of attention.

Neural interactions in working memory explain decreased recall precision and similarity-based feature repulsion

Over the last several years, the study of working memory (WM) for simple visual features (e.g., colors, orientations) has been dominated by perspectives that assume items in WM are stored independently of one another. Evidence has revealed, however, systematic biases in WM recall which suggest that items in WM interact during active maintenance. In the present study, we report two experiments that replicate a repulsion bias between metrically similar colors during active storage in WM. We also observed that metrically similar colors were stored with lower resolution than a unique color held actively in mind at the same time. To account for these effects, we report quantitative simulations of two novel neurodynamical models of WM. In both models, the unique behavioral signatures reported here emerge directly from laterally-inhibitory neural interactions that serve to maintain multiple, distinct neural representations throughout the WM delay period. Simulation results show that the full pattern of empirical findings was only obtained with a model that included an elaborated spatial pathway with sequential encoding of memory display items. We discuss implications of our findings for theories of visual working memory more generally.

Swap errors in visual working memory are fully explained by cue-feature variability

In cue-based recall from working memory, incorrectly reporting features of an uncued item may be referred to as a "swap" error. One account of these errors ascribes them to variability in memory for the cue features leading to erroneous selection of a non-target item, especially if it is similar to the target in the cue-feature dimension. However, alternative accounts of swap errors include cue-independent misbinding, and strategic guessing when the cued item is not in memory. Here we investigated the cause of swap errors by manipulating the variability with which either cue or report features (orientations in Exp 1; motion directions in Exp 2) were encoded. We found that swap errors increased with increasing variability in memory for the cue features, and their changing frequency could be quantitatively predicted based on recall variability when the same feature was used for report. These results are inconsistent with the hypothesis that swaps are a strategic response to forgotten items, and suggest that swap errors could be wholly accounted for by confusions due to cue-dimension variability. In a third experiment we examined whether spatial configuration of memory arrays in tasks with spatial cueing has an influence on swap error frequency. We observed a specific tendency to make swap errors to non-targets located precisely opposite to the cued location, suggesting that stimulus positions are partially encoded in a non-metric format.

Little Support for Discrete Item Limits in Visual Working Memory

Some theorists argue that working memory is limited to a discrete number of items and that additional items are not encoded at all. Adam et al. (2017) presented evidence supporting this hypothesis: Participants reproduced visual features of up to six items in a self-chosen order. After the third or fourth response, error distributions were indistinguishable from guessing. I present four experiments with young adults (each N = 24) reexamining this finding. Experiment 1 presented items slowly and sequentially. Experiment 2 presented them simultaneously but longer than in the experiments of Adam et al. Experiments 3 and 4 exactly replicated one original experiment of Adam et al. All four experiments failed to replicate the evidence for guessing-like error distributions. Modeling data from individuals revealed a mixture of some who do and others who do not produce guessing-like distributions. This heterogeneity increases the credibility of an alternative to the item-limit hypothesis: Some individuals decide to guess on hard trials even when they have weak information in memory.

The role of ensemble average differs in working memory for depth and planar information

The representation of individual planar locations and features stored in working memory can be affected by the average representation. However, less is known about how the average representation affects the short-term storage of depth information. To evaluate the possible different roles of the ensemble average in working memory for planar and depth information, we used mathematical models to fit the data collected from one study on working memory for depth and 12 studies on working memory for planar information. The pattern of recalled depth was well captured by models assuming that there was a probability of reporting the average depth instead of the individual depth, compressing the recalled front-back distance of the stimulus ensemble compared to the perceived distance. However, when modeling the recalled planar information, we found that participants tended to report individual nontarget features when the target was not memorized, and the assumption of reporting average information improves the data fitting only in very few studies. These results provide evidence for our hypothesis that average depth information can be used as a substitution for individual depth information stored in working memory, but for planar visual features, the substitution of target with the average works under a constraint that the average of to-be-remembered features is readily accessible.

Four Internal Inconsistencies in Tversky and Kahneman’s (1992) Cumulative Prospect Theory Article: A Case Study in Ambiguous Theoretical Scope and Ambiguous Parsimony

Scholars heavily rely on theoretical scope as a tool to challenge existing theory. We advocate that scientific discovery could be accelerated if far more effort were invested into also overtly specifying and painstakingly delineating the intended purview of any proposed new theory at the time of its inception. As a case study, we consider Tversky and Kahneman (1992). They motivated their Nobel-Prize-winning cumulative prospect theory with evidence that in each of two studies, roughly half of the participants violated independence, a property required by expected utility theory (EUT). Yet even at the time of inception, new theories may reveal signs of their own limited scope. For example, we show that Tversky and Kahneman’s findings in their own test of loss aversion provide evidence that at least half of their participants violated their theory, in turn, in that study. We highlight a combination of conflicting findings in the original article that make it ambiguous to evaluate both cumulative prospect theory’s scope and its parsimony on the authors’ own evidence. The Tversky and Kahneman article is illustrative of a social and behavioral research culture in which theoretical scope plays an extremely asymmetric role: to call existing theory into question and motivate surrogate proposals.

Modeling mean estimation tasks in within-trial and across-trial contexts

The mean estimation task, which explicitly asks observers to estimate the mean feature value of multiple stimuli, is a fundamental paradigm in research areas such as ensemble coding and cue integration. The current study uses computational models to formalize how observers summarize information in mean estimation tasks. We compare model predictions from our Fidelity-based Integration Model (FIM) and other models on their ability to simulate observed patterns in within-trial weight distribution, across-trial information integration, and set-size effects on mean estimation accuracy. Experiments show non-equal weighting within trials in both sequential and simultaneous mean estimation tasks. Observers implicitly overestimated trial means below the global mean and underestimated trial means above the global mean. Mean estimation performance declined and stabilized with increasing set sizes. FIM successfully simulated all observed patterns, while other models failed. FIM's information sampling structure provides a new way to interpret the capacity limit in visual working memory and sub-sampling strategies. As a model framework, FIM offers task-dependent modeling for various ensemble coding paradigms, facilitating research synthesis across different studies in the literature.

An interference model for visual working memory: Applications to the change detection task

Most studies of visual-working memory employ one of two experimental paradigms: change-detection or continuous-stimulus reproduction. In this study, we extended the Interference Model (IM; Oberauer & Lin, 2017), which was designed for continuous reproduction, to the single-probe change-detection task. In continuous reproduction, participants occasionally report the non-target items instead of the target. The presence of non-target response is predicted by the Interference Model, which relies in part on the interference of non-target items to explain the set-size effect. By presenting a probe matching a non-target item, we can investigate the amount of interference from non-target items in change detection. As predicted by the Interference Model, we observed poorer performance in rejecting a probe matching a non-target item compared to a new probe (i.e., a cost due to intrusions from non-targets). We fitted the IM along with the Variable Precision, the Slot-Averaging, and the Neural-Population model to the data from two change-detection experiments. The models were equipped with a Bayesian decision rule based on the one used in Keshvari, van den Berg, and Ma (2013). The Interference Model and the Neural-Population model successfully predicted the set-size effect and the non-target intrusion cost, whereas the Variable Precision (VP) and Slot-Averaging (SA) models failed to predict the intrusion cost at all. Even with additional assumptions enabling VP and SA to produce intrusion costs, the IM still performed better than the competing models quantitatively.

mixtur: An R package for designing, analysing, and modelling continuous report visual short-term memory studies

Visual short-term memory (vSTM) is often measured via continuous-report tasks whereby participants are presented with stimuli that vary along a continuous dimension (e.g., colour) with the goal of memorising the stimulus features. At test, participants are probed to recall the feature value of one of the memoranda in a continuous manner (e.g., by clicking on a colour wheel). The angular deviation between the participant response and the true feature value provides an estimate of recall precision. Two prominent models of performance on such tasks are the two- and three-component mixture models (Bays et al., Journal of Vision, 9(10), Article 7, 2009; Zhang and Luck, Nature, 453(7192), 233–235, 2008). Both models decompose participant responses into probabilistic mixtures of: (1) responses to the true target value based on a noisy memory representation; (2) random guessing when memory fails. In addition, the three-component model proposes (3) responses to a non-target feature value (i.e., binding errors). Here we report the development of mixtur, an open-source package written for the statistical programming language R that facilitates the fitting of the two- and three-component mixture models to continuous report data. We also conduct simulations to develop recommendations for researchers on trial numbers, set sizes, and memoranda similarity, as well as parameter recovery and model recovery. In the Discussion, we discuss how mixtur can be used to fit the slots and the slots-plus-averaging models, as well as how mixtur can be extended to fit explanatory models of visual short-term memory. It is our hope that mixtur will lower the barrier of entry for utilising mixture modelling.

Working memory capacity of crows and monkeys arises from similar neuronal computations

Complex cognition relies on flexible working memory, which is severely limited in its capacity. The neuronal computations underlying these capacity limits have been extensively studied in humans and in monkeys, resulting in competing theoretical models. We probed the working memory capacity of crows (Corvus corone) in a change detection task, developed for monkeys (Macaca mulatta), while we performed extracellular recordings of the prefrontal-like area nidopallium caudolaterale. We found that neuronal encoding and maintenance of information were affected by item load, in a way that is virtually identical to results obtained from monkey prefrontal cortex. Contemporary neurophysiological models of working memory employ divisive normalization as an important mechanism that may result in the capacity limitation. As these models are usually conceptualized and tested in an exclusively mammalian context, it remains unclear if they fully capture a general concept of working memory or if they are restricted to the mammalian neocortex. Here, we report that carrion crows and macaque monkeys share divisive normalization as a neuronal computation that is in line with mammalian models. This indicates that computational models of working memory developed in the mammalian cortex can also apply to non-cortical associative brain regions of birds.

Mechanisms of feature binding in visual working memory are stable over long delays

The ability to accurately retain the binding between the features of different objects is a critical element of visual working memory. The underlying mechanism can be elucidated by analyzing correlations of response errors in dual-report experiments, in which participants have to report two features of a single item from a previously viewed stimulus array. Results from separate previous studies using different cueing conditions have indicated that location takes a privileged role in mediating binding between other features, in that largely independent response errors have been observed when location was used as a cue, but errors were highly correlated when location was one of the reported features. Earlier results from change detection tasks likewise support such a special role of location, but they also suggest that this role is substantially reduced for longer retention intervals in favor of object-based representation. In the present study, we replicated the findings of previous dual-report tasks with different cueing conditions, using matched stimuli and procedures. Moreover, we show that the observed patterns of error correlations remain qualitatively unchanged with longer retention intervals. Fits with neural population models demonstrate that the behavioral results at long, as well as short, delays are best explained by memory representations in independent feature maps, in which an item's features are bound to each other only via their shared location.

Uncertainty is maintained and used in working memory

What are the contents of working memory? In both behavioral and neural computational models, a working memory representation is typically described by a single number, namely, a point estimate of a stimulus. Here, we asked if people also maintain the uncertainty associated with a memory and if people use this uncertainty in subsequent decisions. We collected data in a two-condition orientation change detection task; while both conditions measured whether people used memory uncertainty, only one required maintaining it. For each condition, we compared an optimal Bayesian observer model, in which the observer uses an accurate representation of uncertainty in their decision, to one in which the observer does not. We find that this “Use Uncertainty” model fits better for all participants in both conditions. In the first condition, this result suggests that people use uncertainty optimally in a working memory task when that uncertainty information is available at the time of decision, confirming earlier results. Critically, the results of the second condition suggest that this uncertainty information was maintained in working memory. We test model variants and find that our conclusions do not depend on our assumptions about the observer's encoding process, inference process, or decision rule. Our results provide evidence that people have uncertainty that reflects their memory precision on an item-specific level, maintain this information over a working memory delay, and use it implicitly in a way consistent with an optimal observer. These results challenge existing computational models of working memory to update their frameworks to represent uncertainty.

Transsaccadic integration relies on a limited memory resource

Saccadic eye movements cause large-scale transformations of the image falling on the retina. Rather than starting visual processing anew after each saccade, the visual system combines post-saccadic information with visual input from before the saccade. Crucially, the relative contribution of each source of information is weighted according to its precision, consistent with principles of optimal integration. We reasoned that, if pre-saccadic input is maintained in a resource-limited store, such as visual working memory, its precision will depend on the number of items stored, as well as their attentional priority. Observers estimated the color of stimuli that changed imperceptibly during a saccade, and we examined where reports fell on the continuum between pre- and post-saccadic values. Bias toward the post-saccadic color increased with the set size of the pre-saccadic display, consistent with an increased weighting of the post-saccadic input as precision of the pre-saccadic representation declined. In a second experiment, we investigated if transsaccadic memory resources are preferentially allocated to attentionally prioritized items. An arrow cue indicated one pre-saccadic item as more likely to be chosen for report. As predicted, valid cues increased response precision and biased responses toward the pre-saccadic color. We conclude that transsaccadic integration relies on a limited memory resource that is flexibly distributed between pre-saccadic stimuli.

Location-independent feature binding in visual working memory for sequentially presented objects

Spatial location is believed to have a privileged role in binding features held in visual working memory. Supporting this view, Pertzov and Husain (Attention, Perception, & Psychophysics, 76(7), 1914–1924, 2014) reported that recall of bindings between visual features was selectively impaired when items were presented sequentially at the same location compared to sequentially at different locations. We replicated their experiment, but additionally tested whether the observed impairment could be explained by perceptual interference during encoding. Participants viewed four oriented bars in highly discriminable colors presented sequentially either at the same or different locations, and after a brief delay were cued with one color to reproduce the associated orientation. When we used the same timing as the original study, we reproduced its key finding of impaired binding memory in the same-location condition. Critically, however, this effect was significantly modulated by the duration of the inter-stimulus interval, and disappeared if memoranda were presented with longer delays between them. In a second experiment, we tested whether the effect generalized to other visual features, namely reporting of colors cued by stimulus shape. While we found performance deficits in the same-location condition, these did not selectively affect binding memory. We argue that the observed effects are best explained by encoding interference, and that memory for feature binding is not necessarily impaired when memoranda share the same location.

Human Representation Learning

The central theme of this review is the dynamic interaction between information selection and learning. We pose a fundamental question about this interaction: How do we learn what features of our experiences are worth learning about? In humans, this process depends on attention and memory, two cognitive functions that together constrain representations of the world to features that are relevant for goal attainment. Recent evidence suggests that the representations shaped by attention and memory are themselves inferred from experience with each task. We review this evidence and place it in the context of work that has explicitly characterized representation learning as statistical inference. We discuss how inference can be scaled to real-world decisions by approximating beliefs based on a small number of experiences. Finally, we highlight some implications of this inference process for human decision-making in social environments.

Stochastic sampling provides a unifying account of visual working memory limits

Research into human working memory limits has been shaped by the competition between different formal models, with a central point of contention being whether internal representations are continuous or discrete. Here we describe a sampling approach derived from principles of neural coding as a framework to understand working memory limits. Reconceptualizing existing models in these terms reveals strong commonalities between seemingly opposing accounts, but also allows us to identify specific points of difference. We show that the discrete versus continuous nature of sampling is not critical to model fits, but that, instead, random variability in sample counts is the key to reproducing human performance in both single- and whole-report tasks. A probabilistic limit on the number of items successfully retrieved is an emergent property of stochastic sampling, requiring no explicit mechanism to enforce it. These findings resolve discrepancies between previous accounts and establish a unified computational framework for working memory that is compatible with neural principles.