Bidirectional influences of information sampling and concept learning.

Abstract

Contemporary models of categorization typically tend to sidestep the problem of how information is initially encoded during decision-making. Instead, a focus of this work has been to investigate how, through selective attention, stimulus representations are contorted such that behaviourally-relevant dimensions are accentuated (or "stretched"), and representations of irrelevant dimensions are ignored (or "compressed"). In high-dimensional real-world environments, it is computationally infeasible to sample all available information, and human decision-makers selectively sample information from sources expected to provide relevant information. To address these and other shortcomings, we develop an active sampling model, Sampling Emergent Attention (SEA), which sequentially and strategically samples information sources until the expected cost of information exceeds the expected benefit. The model specifies the interplay of two components, one involved in determining the expected utility of different information sources and the other in representing knowledge and beliefs about the environment. These two components interact such that knowledge of the world guides information sampling, and what is sampled updates knowledge. Like human decision-makers, the model displays strategic sampling behaviour, such as terminating information search when sufficient information has been sampled and adaptively adjusting the search path in response to previously sampled information. The model also shows human-like failure modes. For example, when information exploitation is prioritized over exploration, the bidirectional influences between information-sampling and learning can lead to the development of beliefs that systematically differ from reality.

Learning and Representation of Hierarchical Concepts in Hippocampus and Prefrontal Cortex

“…The positive relation between mPFC activity and accumulation might be interpreted in light of previous theories on context-dependent memory retrieval in mPFC (Preston and Eichenbaum, 2013): With increasing hierarchical knowledge, the Level 1 feature might increasingly set the context for retrieving the appropriate Level 2 rule. Alternatively, one might consider context (Level 1 feature)-dependent sampling of respectively relevant information in a given trial (Level 2 feature) (Braunlich and Love, 2021) to underlie the accumulation effect in mPFC. Contemporary cognitive models of concept learning (e.g., SUSTAIN, Love et al, 2004;ALCOVE, Kruschke, 1992) mostly operated by accentuating behaviorally relevant stimulus dimensions the same way for every stimulus and would as such have been unable to account for learning problems, such as the present.…”

“…Contemporary cognitive models of concept learning (e.g., SUSTAIN, Love et al, 2004;ALCOVE, Kruschke, 1992) mostly operated by accentuating behaviorally relevant stimulus dimensions the same way for every stimulus and would as such have been unable to account for learning problems, such as the present. Indeed, the assumption of dimension-wide attention is challenged by a new model (SEA; Braunlich and Love, 2021), showing that dynamic and sequential allocation of attention within a trial for active sampling of context-relevant information can likewise account for a range of classical category learning phenomena. Future work might explore the explanatory power of this active sampling model for PFC activation in hierarchical learning problems.…”

As within, so without, as above, so below: Common mechanisms can support between- and within-trial category learning dynamics.

“…ALCOVE, SUSTAIN, AARM, and SEA all fundamentally specify learning as an optimization problem with respect to the observer's goals, but use different mechanisms to solve it. Given that the models make very clear predictions for how dimensions of information are attended over time in order to predict learning (Braunlich & Love, 2021;Galdo et al, 2021;Kruschke, 1992;Love et al, 2004;Mack et al, 2013Mack et al, , 2016, constraining and adjudicating between their respective theoretical assumptions potentially requires insights beyond what behavioral data alone can provide.…”

“…Although SEA’s calculations are driven by predicted utility rather than a theoretical measure of attention, it is worth noting that SEA similarly does not impose a sum-to-constant constraint on its estimates of utility. The model instead implements parsimonious resource expenditure by (a) comparing the predicted utility of sampling a dimension to an expected cost and (b) limiting the depth of forward search when predicting utility (i.e., what Braunlich and Love (2021) refer to as a “mypoic” rather than full preposterior analysis). Through ongoing utility calculations, SEA predicts self-termination when the potential gain of sampling any dimension no longer exceeds the potential cost of time and energy.…”

“…Despite overwhelming evidence that eye-tracking data provide a rich source of information about the timecourse of selective attention during individual decisions, few efforts have been made to extend the logic of categorization models to account for within-trial dynamics (but see Braunlich & Love, 2021). Taking these findings together, we assert that gaze fixations serve as a viable, necessary means for evaluating category learning models in terms of predicted attention allocation.…”