scholarly journals Latent learning, cognitive maps, and curiosity

2021 ◽  
Vol 38 ◽  
pp. 1-7
Author(s):  
Maya Zhe Wang ◽  
Benjamin Y Hayden
2020 ◽  
Author(s):  
Maya Zhe Wang ◽  
Benjamin Y. Hayden

ABSTRACTCuriosity refers to a desire for information that is not driven by immediate strategic or instrumental concerns. Latent earning refers to a form of learning that is not directly driven by standard reinforcement learning processes. We propose that curiosity serves the purpose of motivating latent learning. Thus, while latent learning is often treated as an incidental or passive process, in practice it most often reflects a strong evolved pressure to consume large amounts of information. That large volume of information in turn allows curious decision makers to generate sophisticated representations of the structure of their environment, known as cognitive maps. Cognitive maps facilitate adaptive and flexible behavior while maintaining its adaptivity and flexibility via map updates based on new information. Here we describe data supporting the idea that orbitofrontal cortex (OFC) and dorsal anterior cingulate cortex (dACC) play complementary roles in curiosity-driven learning. Specifically, we propose that (1) OFC tracks the innate value of information and incorporates new information into a detailed cognitive map; and (2) dACC tracks the environmental demands and information availability to then use the cognitive map for guiding behavior.


2020 ◽  
Author(s):  
Ida Momennejad

Memory and planning rely on learning the structure of relationships among experiences. Compact representations of these structures guide flexible behavior in humans and animals. A century after ‘latent learning’ experiments summarized by Tolman, the larger puzzle of cognitive maps remains elusive: how does the brain learn and generalize relational structures? This review focuses on a reinforcement learning (RL) approach to learning compact representations of the structure of states. We review evidence showing that capturing structures as predictive representations updated via replay offers a neurally plausible account of human behavior and the neural representations of predictive cognitive maps. We highlight multi-scale successor representations, prioritized replay, and policy-dependence. These advances call for new directions in studying the entanglement of learning and memory with prediction and planning.


Author(s):  
Márcio Mendonça ◽  
Guilherme Bender Sartori ◽  
Lucas Botoni de Souza ◽  
Giovanni Bruno Marquini Ribeiro

Sign in / Sign up

Export Citation Format

Share Document