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

Kaiyun Li,Xiaolan Fu,Xiaoyan Zhou,Qiufang Fu,Xunwei Sun

doi:10.3389/fpsyg.2016.01017

Abstract

It remains unclear whether probabilistic category learning in the feedback-based weather prediction task (FB-WPT) can be mediated by a non-declarative or procedural learning system. To address this issue, we compared the effects of training time and verbal working memory, which influence the declarative learning system but not the non-declarative learning system, in the FB and paired-associate (PA) WPTs, as the PA task recruits a declarative learning system. The results of Experiment 1 showed that the optimal accuracy in the PA condition was significantly decreased when the training time was reduced from 7 to 3 s, but this did not occur in the FB condition, although shortened training time impaired the acquisition of explicit knowledge in both conditions. The results of Experiment 2 showed that the concurrent working memory task impaired the optimal accuracy and the acquisition of explicit knowledge in the PA condition but did not influence the optimal accuracy or the acquisition of self-insight knowledge in the FB condition. The apparent dissociation results between the FB and PA conditions suggested that a non-declarative or procedural learning system is involved in the FB-WPT and provided new evidence for the multiple-systems theory of human category learning.

Highlights

It remains controversial whether probabilistic category learning can be mediated primarily by a non-declarative or procedural learning system (Poldrack et al, 2001; Newell et al, 2007; Poldrack and Foerde, 2008)
The FB and PA versions of the weather prediction task (WPT) share the common goal of learning the value of probabilistic cues (Dickerson et al, 2011), the PA-WPT is widely accepted to recruit the declarative learning system, whereas whether the feedback-based WPT (FB-WPT) employs a nondeclarative or procedural learning system remains controversial
To investigate whether probabilistic category learning can be dependent on a non-declarative or procedural learning system, we compared the effects of training time and verbal working memory that influence the verbal learning system but not the non-verbal learning system in the FB-WPT and the PA-WPT

Summary

Introduction

It remains controversial whether probabilistic category learning can be mediated primarily by a non-declarative or procedural learning system (Poldrack et al, 2001; Newell et al, 2007; Poldrack and Foerde, 2008). As described in Knowlton et al (1994), in the WPT, participants were given multidimensional stimuli (a set of four tarot cards as cues) and asked to classify those stimuli into one of two categories (rainy or sunny outcome); the cueoutcome associations are probabilistic, and participants can learn them through trial-by-trial feedback. It was found that people with amnesia exhibited normal learning performance on the FB-WPT but expressed poor explicit memory of this task, whereas people with Parkinson’s and Huntington’s disease showed impaired learning of this task but good explicit memory of the task features (Knowlton et al, 1996a,b) These findings provided evidence for the multiple category-learning systems assumption, which assumed that category learning is mediated by one declarative system and one non-declarative or procedural system (Zeithamova and Maddox, 2007). It has been demonstrated that participants reported accurate knowledge of both the task structure and their own judgment processes in the FB-WPT, indicating that the learning is mediated by one explicit declarative learning system (Lagnado et al, 2006)

Methods

Results

Conclusion