ICCM: Learning Processes

Attention-deficit/hyperactivity disorder (ADHD) is characterized by difficulty in acting in a goal-directed manner. While most environments require a sequence of actions for goal attainment, ADHD was never studied in the context of value-based sequence learning. Here, we made use of current advancements in hierarchical reinforcement-learning algorithms to track the internal value and choice policy of individuals with ADHD performing a three-stage sequence learning task. Specifically, 54 participants (28 ADHD, 26 controls) completed a value-based reinforcement-learning task that allowed us to estimate internal action values for each trial and stage using computational modeling. We found attenuated sensitivity to action values in ADHD compared to controls, both in choice and reaction-time variability estimates. Remarkably, this was found only for first-stage actions (i.e., initiatory actions), while for actions performed just before outcome delivery the two groups were strikingly indistinguishable. These results suggest a difficulty in following value estimation for initiatory actions in ADHD.

This is an in-person presentation on July 22, 2024 (11:00 ~ 11:20 CEST).

A fundamental issue in cognitive science concerns the interaction of the cognitive “how” operations, the genetic/memetic “why” processes, and by what means this interaction results in constrained variability and individual differences. This study proposes a single GEVL model that combines complex cognitive mechanisms with a genetic programming approach. The model evolves populations of cognitive agents, with each agent learning by chunking and incorporating LTM and STM stores, as well as attention. The model simulates two different verbal learning tasks: one that investigates the effect of stimulus-response (S-R) similarity on the learning rate; and the other, that examines how the learning time is affected by the change in stimuli presentation times. GEVL’s results are compared to both human data and EPAM – a different verbal learning model that utilises hand-crafted task-specific strategies. The automatically evolved GEVL strategies produced good fit to the human data in both studies, improving on EPAM’s scores by as much as factor of two on some of the pattern similarity conditions. These findings offer further support to the mechanisms proposed by chunking theory, connect them to the evolutionary approach, and make further inroads towards a Unified Theory of Cognition (Newell, 1990).

This is an in-person presentation on July 22, 2024 (10:00 ~ 10:20 CEST).

Model-based adaptive learning systems have successfully improved the efficiency of fact learning in educational practice. Typically, such systems work by keeping track of a learner’s memory processes by measuring behavior during learning, and using this information to tailor the learning process towards the needs of individual learners. Where many adaptive learning systems applied today focus on learning paired associates, we here focus on learning grammar rules based on instances of these general rules. We show that participants’ (N = 42) behavioral responses on instance questions for a rule can be used to infer general performance on other ques- tions associated to that rule, and that we can capture this rule perfor- mance in a single model-based speed of forgetting parameter. These findings could be used to develop and optimize adaptive learning systems that can be used to study general rules from instances.

This is an in-person presentation on July 22, 2024 (10:20 ~ 10:40 CEST).

Many different theories of learning have been developed to account for human performance over time, often accounting for performance at an aggregate level. Understanding performance at an individual level is often more difficult because of multiple different factors—e.g., noise, strategy selection, or change in memory representation—, which are often not ac- counted for in simple learning theories. One approach used to explain the sudden changes in performance that are often observed at the individual level is to integrate change detection algorithms with psychological models. This research has shown that performance at the individual level can be understood not by a single continuous process but instead by segmented portions of multiple processes. Previous research has posited different explanations as to what features drive the inferences of change points. However, no paper has yet compared different explanations’ ability to explain the variance in inferred change points. In this paper, we use a simple model of learning to account for performance in a real-world data set with individuals performing multiple different games that tap into different task attributes (i.e., memory, attention, problem-solving) on the website Luminosity. We then conduct a statistical analysis to determine what drives change points in the dataset. The results here allow for better clarification as to what features are driving the inferences of change points at the individual level.

This is an in-person presentation on July 22, 2024 (10:40 ~ 11:00 CEST).