Task Termination Triggers Spontaneous Removal of Information From Visual Working Memory.

Working memory (WM) is a goal-directed memory system that actively maintains a limited amount of task-relevant information to serve the current goal. By this definition, WM maintenance should be terminated after the goal is accomplished, spontaneously removing no-longer-relevant information from WM. Past studies have failed to provide direct evidence of spontaneous removal of WM content by allowing participants to engage in a strategic reallocation of WM resources to competing information within WM. By contrast, we provide direct neural and behavioral evidence that visual WM content can be largely removed less than 1 s after it becomes obsolete, in the absence of a strategic allocation of resources (total N = 442 adults). These results demonstrate that visual WM is intrinsically a goal-directed system, and spontaneous removal provides a means for capacity-limited WM to keep up with ever-changing demands in a dynamic environment.

A Transient High-dimensional Geometry Affords Stable Conjunctive Subspaces for Efficient Action Selection.

Flexible action selection requires cognitive control mechanisms capable of mapping the same inputs to diverse output actions depending on goals and contexts. How the brain encodes information to enable this capacity remains one of the longstanding and fundamental problems in cognitive neuroscience. From a neural state-space perspective, solving this problem requires a control representation that can disambiguate similar input neural states, making task-critical dimensions separable depending on the context. Moreover, for action selection to be robust and time-invariant, control representations must be stable in time, thereby enabling efficient readout by downstream processing units. Thus, an ideal control representation should leverage geometry and dynamics that maximize the separability and stability of neural trajectories for task computations. Here, using novel EEG decoding methods, we investigated how the geometry and dynamics of control representations constrain flexible action selection in the human brain. Specifically, we tested the hypothesis that encoding a temporally stable conjunctive subspace that integrates stimulus, response, and context (i.e., rule) information in a high-dimensional geometry achieves the separability and stability needed for context-dependent action selection. Human participants performed a task that requires context-dependent action selection based on pre-instructed rules. Participants were cued to respond immediately at varying intervals following stimulus presentation, which forced responses at different states in neural trajectories. We discovered that in the moments before successful responses, there was a transient expansion of representational dimensionality that separated conjunctive subspaces. Further, we found that the dynamics stabilized in the same time window, and that the timing of entry into this stable and high-dimensional state predicted the quality of response selection on individual trials. These results establish the neural geometry and dynamics the human brain needs for flexible control over behavior.

The role of conjunctive representations in prioritizing and selecting planned actions.

For flexible goal-directed behavior, prioritizing and selecting a specific action among multiple candidates are often important. Working memory has long been assumed to play a role in prioritization and planning, while bridging cross-temporal contingencies during action selection. However, studies of working memory have mostly focused on memory for single components of an action plan, such as a rule or a stimulus, rather than management of all of these elements during planning. Therefore, it is not known how post-encoding prioritization and selection operate on the entire profile of representations for prospective actions. Here, we assessed how such control processes unfold over action representations, highlighting the role of conjunctive representations that nonlinearly integrate task-relevant features during maintenance and prioritization of action plans. For each trial, participants prepared two independent rule-based actions simultaneously, then they were retro-cued to select one as their response. Prior to the start of the trial, one rule-based action was randomly assigned to be high priority by cueing that it was more likely to be tested. We found that both full action plans were maintained as conjunctive representations during action preparation, regardless of priority. However, during output selection, the conjunctive representation of the high-priority action plan was more enhanced and readily selected as an output. Furthermore, the strength of the high-priority conjunctive representation was associated with behavioral interference when the low-priority action was tested. Thus, multiple alternate upcoming actions were maintained as integrated representations and served as the target of post-encoding attentional selection mechanisms to prioritize and select an action from within working memory.

The Role of Conjunctive Representations in Stopping Actions.

Action selection appears to rely on conjunctive representations that nonlinearly integrate task-relevant features. Here, we tested a corollary of this hypothesis: that such representations are also intricately involved during attempts to stop an action-a key aspect of action regulation. We tracked both conjunctive representations and those of constituent rule, stimulus, or response features through trial-by-trial representational similarity analysis of the electroencephalogram signal in a combined rule-selection and stop-signal paradigm. Across two experiments with student participants (N = 57), we found (a) that the strength of decoded conjunctive representations prior to the stop signal uniquely predicted trial-by-trial stopping success (Experiment 1) and (b) that these representations were selectively suppressed following the onset of the stop signal (Experiments 1 and 2). We conclude that conjunctive representations are key to successful action execution and therefore need to be suppressed when an intended action is no longer appropriate.

The dimensionality of neural representations for control.

Cognitive control allows us to think and behave flexibly based on our context and goals. At the heart of theories of cognitive control is a control representation that enables the same input to produce different outputs contingent on contextual factors. In this review, we focus on an important property of the control representation's neural code: its representational dimensionality. Dimensionality of a neural representation balances a basic separability/generalizability trade-off in neural computation. We will discuss the implications of this trade-off for cognitive control. We will then briefly review current neuroscience findings regarding the dimensionality of control representations in the brain, particularly the prefrontal cortex. We conclude by highlighting open questions and crucial directions for future research.

Author Correction: EEG Decoding Reveals the Strength and Temporal Dynamics of Goal-Relevant Representations.

An amendment to this paper has been published and can be accessed via a link at the top of the paper.

Conjunctive representations that integrate stimuli, responses, and rules are critical for action selection.

People can use abstract rules to flexibly configure and select actions for specific situations, yet how exactly rules shape actions toward specific sensory and/or motor requirements remains unclear. Both research from animal models and human-level theories of action control point to the role of highly integrated, conjunctive representations, sometimes referred to as event files. These representations are thought to combine rules with other, goal-relevant sensory and motor features in a nonlinear manner and represent a necessary condition for action selection. However, so far, no methods exist to track such representations in humans during action selection with adequate temporal resolution. Here, we applied time-resolved representational similarity analysis to the spectral-temporal profiles of electroencephalography signals while participants performed a cued, rule-based action selection task. In two experiments, we found that conjunctive representations were active throughout the entire selection period and were functionally dissociable from the representation of constituent features. Specifically, the strength of conjunctions was a highly robust predictor of trial-by-trial variability in response times and was selectively related to an important behavioral indicator of conjunctive representations, the so-called partial-overlap priming pattern. These results provide direct evidence for conjunctive representations as critical precursors of action selection in humans.

The Effects of Three-Dimensional Context on Shape Perception.

Humans have a unique ability to perceive shape in different ways. Although we naturally estimate objective (physical) shape in our daily interactions with the world, we are also capable of estimating projective (retinal) shape, especially when attempting to accurately draw objects and scenes. In four experiments, we demonstrated robust effects of 3D context on shape perception. Using a binocular stereo paradigm, we presented rectangular surfaces of varying widths alone or embedded in a polyhedron. We investigated how context, judgment type, and angle affected width estimates. We found that the presence of even a small amount of 3D context aids objective judgments but hinders projective judgments, whereas a lack of context had the opposite effect. Context facilitated objective shape assessments by improving estimates of surface orientation. These results demonstrate that the typical presence of 3D context aids shape perception (shape constancy) while simultaneously making the projective judgments necessary for realistic drawing more difficult.

Does conflict resolution rely on working memory?

Theoretical considerations and results from individual differences studies suggest that working memory and conflict resolution are interrelated functions. Yet, there is little direct evidence suggesting that they actually share common cognitive resources. To study how overcoming conflict influences the maintenance of working memory representations and vice versa, we conducted 4 experiments using a dual-task paradigm in which both working memory load and level of conflict were independently manipulated. Participants performed an auditory Stroop task ("high" or "low" spoken in high/low pitch), which was presented during the retention period of a visual change detection task (Experiments 1-4) or simultaneously with the working memory encoding phase (Experiment 2-4). Across the 4 experiments, we found no consistent interaction between level of conflict and working memory load on working memory performance, although there was evidence in 2 of the 4 experiments for a small effect on auditory Stroop accuracy (but not on response times). These findings present at best weak evidence for the hypothesis that the maintenance of task goals in working memory is critical for successful conflict resolution. (PsycInfo Database Record (c) 2021 APA, all rights reserved).

Balancing model-based and memory-free action selection under competitive pressure.

In competitive situations, winning depends on selecting actions that surprise the opponent. Such unpredictable action can be generated based on representations of the opponent's strategy and choice history (model-based counter-prediction) or by choosing actions in a memory-free, stochastic manner. Across five different experiments using a variant of a matching-pennies game with simulated and human opponents we found that people toggle between these two strategies, using model-based selection when recent wins signal the appropriateness of the current model, but reverting to stochastic selection following losses. Also, after wins, feedback-related, mid-frontal EEG activity reflected information about the opponent's global and local strategy, and predicted upcoming choices. After losses, this activity was nearly absent-indicating that the internal model is suppressed after negative feedback. We suggest that the mixed-strategy approach allows negotiating two conflicting goals: 1) exploiting the opponent's deviations from randomness while 2) remaining unpredictable for the opponent.

EEG Decoding Reveals the Strength and Temporal Dynamics of Goal-Relevant Representations.

Models of action control assume that attentional control settings regulate the processing of lower-level stimulus/response representations. Yet, little is known about how exactly control and sensory/response representations relate to each other to produce goal-directed behavior. Addressing this question requires time-resolved information about the strength of the different, potentially overlapping representations, on a trial-by-trial basis. Using a cued task-switching paradigm, we show that information about relevant representations can be extracted through decoding analyses from the scalp electrophysiological signal (EEG) with high temporal resolution. Peaks in representational strength-indexed through decoding accuracy-proceeded from superficial task cues, to stimulus locations, to features/responses. In addition, attentional-set representations were prominent throughout almost the entire processing cascade. Trial-by-trial analyses provided detailed information about when and to what degree different representations predict performance, with attentional settings emerging as a strong and consistent predictor of within-individual and across-individual variability in performance. Also, the strength of attentional sets was related to target representations early in the post-stimulus period and to feature/response representations at a later period, suggesting control of successive, lower-level representations in a concurrent manner. These results demonstrate a powerful approach towards uncovering different stages of information processing and their relative importance for performance.

Decoding hierarchical control of sequential behavior in oscillatory EEG activity.

Despite strong theoretical reasons for assuming that abstract representations organize complex action sequences in terms of subplans (chunks) and sequential positions, we lack methods to directly track such content-independent, hierarchical representations in humans. We applied time-resolved, multivariate decoding analysis to the pattern of rhythmic EEG activity that was registered while participants planned and executed individual elements from pre-learned, structured sequences. Across three experiments, the theta and alpha-band activity coded basic elements and abstract control representations, in particular, the ordinal position of basic elements, but also the identity and position of chunks. Further, a robust representation of higher level, chunk identity information was only found in individuals with above-median working memory capacity, potentially providing a neural-level explanation for working-memory differences in sequential performance. Our results suggest that by decoding oscillatory activity we can track how the cognitive system traverses through the states of a hierarchical control structure.

The Nature of Task Set Representations in Working Memory.

Selection and preparation of action plans (task sets) is often assumed to occur in working memory (WM). Yet, the absence of consistent evidence that WM capacity and task selection efficiency is correlated raises questions about the functional relationship between these two aspects of executive control. We used the EEG-derived contralateral delay activity (CDA) to index the WM load of task sets. In Experiment 1, we found a CDA set size effect (2 vs. 4 stimulus-response [S-R] rules) for high-WM, but not for low-WM, individuals when S-R sets were novel. In contrast, when only four task sets were presented throughout the experiment, we observed a sustained yet set size-independent use of WM for high-WM participants. Moreover, Experiment 2 showed an increase of the CDA in situations with task conflict, and this effect was larger the more that participants experienced RT conflict effects. Combined, these results indicate that even highly familiar S-R settings are maintained in WM, albeit in a compressed manner, presumably through cues to long-term memory representations. Finally, participants with low-WM capacity represented even familiar tasks in a load-dependent manner, suggesting that the establishment of effective retrieval structures itself is a capacity-limited process.

Dynamics of task-set carry-over: evidence from eye-movement analyses.

Trial-to-trial carry-over of task sets (i.e., task-set inertia) is often considered as a primary reason for task-switch costs. Yet, we know little about the dynamics of such carry-over effects, in particular how much they are driven by the most recent trial rather than characterized by a more continuous memory gradient. Using eye-tracking, we examined in a 3-task, switching paradigm whether there is a greater probability of non-target fixations to stimuli associated with the previously relevant attentional set than to those associated with the less-recent set. Indeed, we found strong evidence for more interference (expressed in terms of non-target fixations) from recent than from less-recent tasks and that in particular the interference from pre-switch trials contributed substantially to the overall pattern of response-time switch costs. Moreover, task-set carry-over was dominated by the most-recent trial when subjects could expect task repetitions (with a 33 % switch rate). In comparison, when tasks were selected randomly (with a 66 % switch rate), interference from the most recent trial decreased, whereas interference from less-recent trials increased. In sum, carry-over interference dynamics were characterized both by a gradual recency gradient and expectations about task-transition probabilities. Beyond that, there was little evidence for a unique role of the most-recent trial.

Control of task sequences: what is the role of language?

It is almost a truism that language aids serial-order control through self-cuing of upcoming sequential elements. We measured speech onset latencies as subjects performed hierarchically organized task sequences while "thinking aloud" each task label. Surprisingly, speech onset latencies and response times (RTs) were highly synchronized, a pattern that is not consistent with the hypothesis that speaking aids proactive retrieval of upcoming sequential elements during serial-order control. We also found that when instructed to do so, subjects were able to speak task labels prior to presentation of response-relevant stimuli and that this substantially reduced RT signatures of retrieval-however, at the cost of more sequencing errors. Thus, while proactive retrieval is possible in principle, in natural situations it seems to be prevented through a strong "gestalt-like" tendency to synchronize speech and action. We suggest that this tendency may support context updating rather than proactive control.