LaPsyDÉ

Testing the corrective assumption of dual process theory in reasoning

Supervised by Wim de Neys

Abstract :

Dual-process theories of reasoning have become widely recognized as an explanation for various phenomena, such as thinking biases, moral or cooperative reasoning. Dual-process theory conceives human thinking as the interaction of a fast, more automatic, intuitive system (System 1) and a slower, controlled, more deliberative one (System 2). Arguably, the most dominant view on dual processes is the default-interventionist model. This posits a serial interaction between the two systems. When someone is faced with a reasoning problem, initially a System 1 intuitive response is formed. Then, afterwards, System 2 might get engaged in the process. Prominent dual-process theorists argue that reasoning bias occurs as a result of erroneous System 1 intuition. System 1 is thought to be able to generate responses based on "heuristic" cues, such as stereotypes - and cannot account for logico-mathematical principles. Despite its huge recognition, this theory comes with an untested assumption: the corrective (time-course) assumption. This posits that in cases when heuristic cues are in conflict with logico-mathematical principles, System 2 needs to engage in order to correct initially formed System 1 intuitions, and form a judgement based on logical principles. Testing this assumption is inevitably important and the central question of this thesis. In Study 1, I used four modified versions of the two-response paradigm to test the corrective assumption with two different classical reasoning problems (base rate problems, syllogistic reasoning). In this paradigm, people are presented with the same problem twice. First, they are asked to give an initial, very quick response. After, they are presented with the same problem again and asked to give a final response without any constraints. To make sure that the initial response is really intuitive, we applied four different procedures: instructions, concurrent load, response deadline and load plus deadline. Dual process theory predicts that logically correct responses appear only at the final response stage. Surprisingly, I found that the majority of people who gave the logically correct response in the final response stage already gave it form the beginning. This effect was found to be consistent among all experimental procedures and both reasoning problems. In Study 2, I tried to test the same assumption, with a different -harder- reasoning problem, the bat-and-ball problem. Interestingly, I ran 7 experiments with the two-response paradigm and consistently found that correct reasoners are often able to generate the correct response from the beginning, so-to-say, intuitively. These results forced me to revise the default-interventionist framework and propose the hybrid dual process model. This model now argues that System 1 generates two kinds of intuitive responses one of which is based on mathematico-logical principles. These responses are generated with unequal strength - the one which gains the more strength will be given as the initial response. In Study 3, I directly tested predictions derived from this model. In Study 4, I further developed the hybrid model by testing the changes in the strength of intuitive responses over time. In Study 5, I started to test the hybrid model's domain generality, and test if I find similar patterns of responses when people are faced with moral dilemmas. In Study 6, I used EEG to search for the neural correlates of early logical processing in reasoning. Overall, this thesis found evidence that forces us to revise the traditional dual process view on human reasoning.