Project Details
Projekt
Predictive processing during task decisions and task performance
Applicants
Professor Dr. Martin Butz; Professor Dr. David Dignath
Subject Area
General, Cognitive and Mathematical Psychology
Term
since 2026
Project identifier
Deutsche Forschungsgemeinschaft (DFG) - Project number 564829065
In everyday life we are continuously confronted with a variety of potential tasks. While there is much known about when, how, and why our brain struggles with parallel processing and task switching, much less is known about how we choose tasks. Why do we get so easily distracted or feel the urge to do something else, particularly when the current task is challenging? Which aspects critically influence our task choices? Which computational processes are unfolding while we choose tasks and while we execute them? To address these questions, we will focus on a recently developed type of voluntary task switching (VTS) paradigm, termed the performance-based decision (PBD) paradigm. In PBD, tasks slowly get harder when they are executed repeatedly. This elicits the tendency to switch the task. The paradigm motivates switches by information-theoretic manipulations (i.e., changes in the signal-to-noise ratio during task performance). PBD experiments thus naturally offer themselves to be modeled by an information-theoretic model. Our goal is to develop a better mechanistic understanding of why and how we decide to switch tasks by means of such a model. We propose to develop a normative, active inference-based predictive processing model of task choices and actual task performance. The model’s formal goal is to minimize (upper bounded) surprise, including response errors. This formalization directly leads to a quantification of benefits, when successfully solving the task, and costs, or efforts, while solving it. We will consider three cost components: Performance effort quantifies the effort to solve one task, that is, to infer the correct action. Task switching effort quantifies the effort to refocus onto the other task. Finally, planning effort refers to the extra effort to consider solving the next actual task before inferring the task choice. To examine how these efforts influence our task choices, we will test manipulations that can be linked to each of these effort costs in three work packages. A first model implementation will account for published PBD data. It will then be put to the test and further scrutinized using the gathered novel data. By means of our experimental approach, we expect to reveal core factors that influence task choices. By means of the modeling approach, we expect to shed light on how our brain mechanistically computes behavioral benefits and costs and how it effectively combines them to generate behavior. In a final integrative work package, we will relate the identified mechanisms and influencing factors and assess if and to which extent working memory capacity influences them. Our project thus will offer normative, mechanistic explanations to the questions of why and how our brain chooses the next task to pursue.
DFG Programme
Research Units
