engleski

Chronometry and meta-reasoning in a modified Cognitive Reflection Test

The Cognitive Reflection Test (CRT) has become common measurement of analytical thinking and rationality. The original version contains three problems of similar structure. The tasks are designed to induce heuristic-biased thinking which commonly leads to the wrong answer which needs to be overridden. The aim of this study was to introduce additional three items and modify the CRT by including chronometric and metacognitive measures of confidence. Participants (N=70) were presented with a problem and four alternatives amongst which one was the correct response, one was based on a heuristic, one was a filler incorrect response and the final alternative was “none of above”. Their task was to respond, as fast as possible, by pressing the key corresponding to their preferred response. After each response, participants gave a judgment of confidence on a seven-point scale. Item-level analysis were conducted in order to compare correct and heuristic responses. There was a total of 28.86% correct and 52.86% of heuristic responses (18.28% filler and “none of the above). Two 2 (heuristic vs analytic response) by 6 (CRT item) ANOVAs were conducted on response times and confidence ratings. The response time analysis revealed a significant main effect of response type (F(1, 327) = 5.60, p < .05, np2 = .02) with slightly faster responses for heuristic responses. However, this was qualified by a significant interaction effect (F(5, 327) = 3.50, p < .01, np2 = .05). The only significant difference in response times between correct and heuristic responses was observed in the Bat and ball problem. The confidence ratings analysis revealed a significant main effect of response type (F(1, 327) = 11.62, p <.01, np2 = .03) as well as CRT problem (F(5, 327) = 3.84, p < 01, np2 = .06), with both effects qualified by a significant interaction (F(5, 327) = 3.05, p < 05, np2 = .04). Participants were generally more confident when providing correct responses, but this was the case for only three of the six problems. Additionally correlation analysis revealed significant negative correlations between response times and confidence ratings for both correct (r(115) = -.29, p < .01) and heuristic responses (r(220) = -.35, p < .01). Shorter response times were accompanied by higher confidence. The results show there is significant variability between the various problems of the modified CRT which will be discussed in light of recent development within the dual-process approach to reasoning.

Cognitive Reflection Test, meta-reasoning, metacognition, heuristic bias, analytical thinking

nije evidentirano