Paired-Associate and Feedback-Based Weather Prediction Tasks Support Multiple Category Learning Systems.

Area of Science:

• Cognitive psychology
• Neuroscience
• Learning sciences

Background:

• The neural basis of category learning is debated, with questions about the involvement of declarative versus non-declarative systems.
• The feedback-based weather prediction task (FB-WPT) is often used to study category learning, but its underlying mechanisms remain unclear.
• Declarative learning is influenced by factors like training time and working memory, while non-declarative learning is less affected.

Purpose of the Study:

• To investigate whether probabilistic category learning in the FB-WPT relies on non-declarative or procedural learning systems.
• To differentiate the learning mechanisms of the FB-WPT from those of the paired-associate (PA) WPT, which engages declarative learning.
• To examine the impact of training time and verbal working memory on performance in both FB and PA tasks.

Main Methods:

• Experiment 1 manipulated training time (7s vs. 3s) in both FB and PA tasks.
• Experiment 2 introduced a concurrent verbal working memory task during FB and PA learning.
• Performance was assessed by optimal accuracy and the acquisition of explicit knowledge.

Main Results:

• Reduced training time impaired optimal accuracy in the PA task but not the FB task.
• A concurrent working memory task hindered performance in the PA task but not the FB task.
• Explicit knowledge acquisition was impaired by reduced training time and working memory load in both tasks, but self-insight knowledge in the FB task remained unaffected.

Conclusions:

• The dissociation observed between the FB and PA tasks suggests that the FB-WPT involves non-declarative or procedural learning.
• These findings provide evidence for the multiple-systems theory of human category learning.
• Different tasks recruit distinct learning systems, highlighting the complexity of human learning.