Collaborative discovery in a simple reasoning task

  • Authors:
  • Kazuhisa Miwa

  • Affiliations:
  • Graduate School of Information Science, Nagoya University, Furo-cho, Nagoya-shi 464-8601, Japan

  • Venue:
  • Cognitive Systems Research
  • Year:
  • 2004

Quantified Score

Hi-index 0.00

Visualization

Abstract

In psychological experimental studies on scientific discovery, discussions have been made on the effects of collaborative discovery using simple experimental tasks, such as Wason's 2-4-6 task. Generally speaking, however, these studies have not explained the prominent effects of multiple subjects collaboratively discovering a target. In the present study, we identify situations in which the effects of collaboration emerge, and why they emerge, by combining a computer simulation method that employs our computational model as a cognitive simulator with a method based on laboratory studies. We basically control two factors, i.e. the hypothesis-testing strategy used by the subjects and the nature of the target that the subjects are required to find, both of which have been identified by previous psychological studies as key factors determining the subjects' performance. The computer simulations show that the performance of combined two systems in collaborative discovery exceeds that of each system in independent discovery, but only when the two systems try to find a target having the nature of generality by repeatedly conducting a positive test. This finding is also confirmed by psychological experiments designed to verify the computer simulations. Moreover, through a theoretical analysis, we show that this effect of collaboration is provided by the emergence of negative tests from the interaction of the two systems (or humans) repeating only a positive test.