Cognitive Control as a Multivariate Optimization Problem

Toward the Emergence of Intelligent Control: Episodic Generalization and Optimization

Human cognition is unique in its ability to perform a wide range of tasks and to learn new tasks quickly. Both abilities have long been associated with the acquisition of knowledge that can generalize across tasks and the flexible use of that knowledge to execute goal-directed behavior. We investigate how this emerges in a neural network by describing and testing the Episodic Generalization and Optimization (EGO) framework. The framework consists of an episodic memory module, which rapidly learns relationships between stimuli; a semantic pathway, which more slowly learns how stimuli map to responses; and a recurrent context module, which maintains a representation of task-relevant context information, integrates this over time, and uses it both to recall context-relevant memories (in episodic memory) and to bias processing in favor of context-relevant features and responses (in the semantic pathway). We use the framework to address empirical phenomena across reinforcement learning, event segmentation, and category learning, showing in simulations that the same set of underlying mechanisms accounts for human performance in all three domains. The results demonstrate how the components of the EGO framework can efficiently learn knowledge that can be flexibly generalized across tasks, furthering our understanding of how humans can quickly learn how to perform a wide range of tasks-a capability that is fundamental to human intelligence.

An experimental manipulation of the value of effort

People who take on challenges and persevere longer are more likely to succeed in life. But individuals often avoid exerting effort, and there is limited experimental research investigating whether we can learn to value effort. We developed a paradigm to test the hypothesis that people can learn to value effort and will seek effortful challenges if directly incentivized to do so. We also dissociate the effects of rewarding people for choosing effortful challenges and performing well. The results provide limited evidence that rewarding effort increased people's willingness to choose harder tasks when rewards were no longer offered (near transfer). There was also mixed evidence that rewarding effort increased willingness to choose harder tasks in another unrelated and unrewarded task (far transfer). These heterogeneous results highlight the need for further research to understand when this paradigm may be the most effective for increasing and generalizing the value of effort.

Orthogonal neural encoding of targets and distractors supports multivariate cognitive control

The complex challenges of our mental life require us to coordinate multiple forms of neural information processing. Recent behavioural studies have found that people can coordinate multiple forms of attention, but the underlying neural control process remains obscure. We hypothesized that the brain implements multivariate control by independently monitoring feature-specific difficulty and independently prioritizing feature-specific processing. During functional MRI, participants performed a parametric conflict task that separately tags target and distractor processing. Consistent with feature-specific monitoring, univariate analyses revealed spatially segregated encoding of target and distractor difficulty in the dorsal anterior cingulate cortex. Consistent with feature-specific attentional priority, our encoding geometry analysis revealed overlapping but orthogonal representations of target and distractor coherence in the intraparietal sulcus. Coherence representations were mediated by control demands and aligned with both performance and frontoparietal activity, consistent with top-down attention. Together, these findings provide evidence for the neural geometry necessary to coordinate multivariate cognitive control.

Control adjustment costs limit goal flexibility: Empirical evidence and a computational account

A cornerstone of human intelligence is the ability to flexibly adjust our cognition and behavior as our goals change. For instance, achieving some goals requires efficiency, while others require caution. Adapting to these changing goals require corresponding adjustments in cognitive control (e.g., levels of attention, response thresholds). However, adjusting our control to meet new goals comes at a cost: we are better at achieving a goal in isolation than when transitioning between goals. The source of these control adjustment costs remains poorly understood, and the bulk of our understanding of such costs comes from settings in which participants transition between discrete task sets, rather than performance goals. Across four experiments, we show that adjustments in continuous control states incur a performance cost, and that a dynamical systems model can explain the source of these costs. Participants performed a single cognitively demanding task under varying performance goals (e.g., to be fast or to be accurate). We modeled control allocation to include a dynamic process of adjusting from one's current control state to a target state for a given performance goal. By incorporating inertia into this adjustment process, our model accounts for our empirical findings that people under-shoot their target control state more (i.e., exhibit larger adjustment costs) when (a) goals switch rather than remain fixed over a block (Study 1); (b) target control states are more distant from one another (Study 2); (c) less time is given to adjust to the new goal (Study 3); and (d) when anticipating having to switch goals more frequently (Study 4). Our findings characterize the costs of adjusting control to meet changing goals, and show that these costs can emerge directly from cognitive control dynamics. In so doing, they shed new light on the sources of and constraints on flexibility in human goal-directed behavior.

Humans reconfigure target and distractor processing to address distinct task demands

When faced with distraction, we can focus more on goal-relevant information (targets) or focus less on goal-conflicting information (distractors). How people use cognitive control to distribute attention across targets and distractors remains unclear. We address this question by developing a novel Parametric Attentional Control Task that can "tag" participants' sensitivity to target and distractor information. We use these precise measures of attention to develop a novel process model that can explain how participants control attention toward targets and distractors. Across three experiments, we find that participants met the demands of this task by independently controlling their processing of target and distractor information, exhibiting distinct adaptations to manipulations of incentives and conflict. Whereas incentives preferentially led to target enhancement, conflict in the previous trial preferentially led to distractor suppression. These distinct drivers of control altered sensitivity to targets and distractors early in the trial, promptly followed by reactive reconfiguration toward task-appropriate feature sensitivity. To provide a process-level account of these empirical findings, we develop a novel neural network model of evidence accumulation with attractor dynamics over feature weights that reconfigure target and distractor processing. These results provide a computational account of control reconfiguration that provides new insights into how multivariate attentional signals are optimized to achieve task goals. (PsycInfo Database Record (c) 2024 APA, all rights reserved).

Motivational context determines the impact of aversive outcomes on mental effort allocation

It is well known that people will exert effort on a task if sufficiently motivated, but how they distribute these efforts across different strategies (e.g., efficiency vs. caution) remains uncertain. Past work has shown that people invest effort differently for potential positive outcomes (rewards) versus potential negative outcomes (penalties). However, this research failed to account for differences in the context in which negative outcomes motivate someone - either as punishment or reinforcement. It is therefore unclear whether effort profiles differ as a function of outcome valence, motivational context, or both. Using computational modeling and our novel Multi-Incentive Control Task, we show that the influence of aversive outcomes on one's effort profile is entirely determined by their motivational context. Participants (N:91) favored increased caution in response to larger penalties for incorrect responses, and favored increased efficiency in response to larger reinforcement for correct responses, whether positively or negatively incentivized.

Policy abstraction as a predictor of cognitive effort avoidance

Consistent evidence has established that people avoid cognitively effortful tasks. However, the features that make a task cognitively effortful are still not well understood. Multiple hypotheses have been proposed regarding which task demands underlie cognitive effort costs, such as time-on-task, error likelihood, and the general engagement of cognitive control. In this study, we test the novel hypothesis that tasks requiring behavior according to higher degrees of policy abstraction are experienced as more effortful. Accordingly, policy abstraction, operationalized as the levels of contextual contingency required by task rules, drives task avoidance over and above the effects of task performance, such as time-on-task or error likelihood. To test this hypothesis, we combined two previously established cognitive control tasks that parametrically manipulated policy abstraction with the demand selection task procedure. The design of these tasks allowed us to test whether people avoided tasks with higher order policy abstraction while controlling for the contribution of factors such as time-on-task and expected error rate (ER). Consistent with our hypothesis, we observed that policy abstraction was the strongest predictor of cognitive effort choices, followed by ER. This was evident across both studies and in a within-subject cross-study analysis. These results establish at least one task feature independent of performance, which is predictive of task avoidance behavior. We interpret these results within an opportunity cost framework for understanding aversive experiences of cognitive effort while performing a task. (PsycInfo Database Record (c) 2023 APA, all rights reserved).

Feeling and deciding: Subjective experiences rather than objective factors drive the decision to invest cognitive control

When presented with the choice to invest cognitive control in a task, several signals are monitored to reach a decision. Leading theoretical frameworks argued that the investment of cognitive control is determined by a cost-benefit computation. However, previous accounts remained silent on the potential role of subjective experience in this computation. We experience confidence when giving an answer, feel the excitement of an anticipated reward, and reflect on how much effort is required for successful task performance. Two questions are investigated in the present work: how objective task parameters give rise to subjective experience and whether these drive the decision to allocate cognitive control. To this end, we designed a task in which we manipulated three objective parameters in the same sequence of events (stimulus uncertainty, physical effort, and reward prediction error). We asked participants to report their subjective experiences associated with these manipulations: confidence, subjective physical effort, and reward satisfaction. At the end of each trial, participants indicated whether they wanted to repeat that trial on the next day. In response to the first question, we demonstrate that subjective ratings are reliable and selective. Subjective experiences closely mirrored their objective manipulations. In response to the second question, we demonstrate that subjective experiences provide a better fit for the decisions on future control investments. While objective task parameters are considered when deciding, they do not always produce the expected changes in subjective experience, and when dissociations occur, it is the subjective experience that better explains the decision to allocate cognitive control.

Developmental changes in exploration resemble stochastic optimization

Human development is often described as a 'cooling off' process, analogous to stochastic optimization algorithms that implement a gradual reduction in randomness over time. Yet there is ambiguity in how to interpret this analogy, due to a lack of concrete empirical comparisons. Using data from n = 281 participants ages 5 to 55, we show that cooling off does not only apply to the single dimension of randomness. Rather, human development resembles an optimization process of multiple learning parameters, for example, reward generalization, uncertainty-directed exploration and random temperature. Rapid changes in parameters occur during childhood, but these changes plateau and converge to efficient values in adulthood. We show that while the developmental trajectory of human parameters is strikingly similar to several stochastic optimization algorithms, there are important differences in convergence. None of the optimization algorithms tested were able to discover reliably better regions of the strategy space than adult participants on this task.

Learning when effort matters: neural dynamics underlying updating and adaptation to changes in performance efficacy

To determine how much cognitive control to invest in a task, people need to consider whether exerting control matters for obtaining rewards. In particular, they need to account for the efficacy of their performance-the degree to which rewards are determined by performance or by independent factors. Yet it remains unclear how people learn about their performance efficacy in an environment. Here we combined computational modeling with measures of task performance and EEG, to provide a mechanistic account of how people (i) learn and update efficacy expectations in a changing environment and (ii) proactively adjust control allocation based on current efficacy expectations. Across 2 studies, subjects performed an incentivized cognitive control task while their performance efficacy (the likelihood that rewards are performance-contingent or random) varied over time. We show that people update their efficacy beliefs based on prediction errors-leveraging similar neural and computational substrates as those that underpin reward learning-and adjust how much control they allocate according to these beliefs. Using computational modeling, we show that these control adjustments reflect changes in information processing, rather than the speed-accuracy tradeoff. These findings demonstrate the neurocomputational mechanism through which people learn how worthwhile their cognitive control is.