Distinctive Delta and Theta Responses in Deductive and Probabilistic Reasoning

Abstract

Introduction: The neural substrates of reasoning, a cognitive ability we use constantly in daily life, are still unclear. Reasoning can be divided into two types according to how the inference process works and the certainty of the conclusions. In deductive reasoning, certain conclusions are drawn from premises by applying the rules of logic. On the other hand, in probabilistic reasoning, possible conclusions are drawn by interpreting the semantic content of arguments.

Methods: We examined event-related oscillations associated with deductive and probabilistic reasoning. To better represent the natural use of reasoning, we adopted a design that required participants to choose what type of reasoning they would use. Twenty healthy participants judged the truth values of alternative conclusion propositions following two premises while the EEG was being recorded. We then analyzed event-related delta and theta power and phase-locking induced under two different conditions.

Results: We found that the reaction time was shorter and the accuracy rate was higher in deductive reasoning than in probabilistic reasoning. High delta and theta power in the temporoparietal, parietal, and occipital regions of the brain were observed in deductive reasoning. As for the probabilistic reasoning, prolonged delta response in the right hemisphere and high frontal theta phase-locking were noted.

Conclusion: Our results suggest that the electrophysiological signatures of the two types of reasoning have distinct characteristics. There are significant differences in the delta and theta responses that are associated with deductive and probabilistic reasoning. Although our findings suggest that deductive and probabilistic reasoning have different neural substrates, consistent with most of the studies in the literature, there is not yet enough evidence to make a comprehensive claim on the subject. There is a need to diversify the growing literature on deductive and probabilistic reasoning with different methods and experimental paradigms.

Keywords: conditional reasoning; deductive reasoning; electroencephalography; event‐related oscillations; probabilistic reasoning.

FIGURE 1

An example of an inference that was presented in the experiment. A premise pair followed by four alternative conclusion propositions is shown.

FIGURE 2

The experiment flow. After each premise pair (P), four alternative conclusion propositions (C) were randomly displayed. The back arrow indicates that other conclusions were shown before moving on to the new premise pair. Participants were instructed to respond when fixation plus appeared. The fixation plus disappeared when they responded, if they did not respond, it remained for a maximum of 2000 ms.

FIGURE 3

Behavioral data results. (a) The reaction time in DR (M = 479, SD = 25.3) and PR (M = 568, SD = 30.3), and (b) the number of correct answers in DR (M = 75.4, SD = 0.919) and PR (M = 68.4, SD = 0.939). Each small dot represents the values of the participants. Big red dots represent the mean value. Bold black lines in the middle of the boxes represent the median value.

FIGURE 4

Event‐related delta power results. (a) Topographic distributions of DR (top) and PR (bottom) for early and late time windows. The red color indicates high power, and the blue color indicates low power. (b) Reasoning type and location interaction in the early time window (p < 0.01). (c) The grand average figures of event‐related delta power analysis (1–3.5 Hz) for TP locations in the time‐frequency domain during DR (top) and PR (bottom). Colors are coded in the same way as section A. (d) Reasoning type and hemisphere interaction in the late time window (p < 0.05). (e) Comparison of reasoning types in early and late time windows (p < 0.05). Asterisks indicate significant results.

FIGURE 5

Event‐related theta power results. (a) Theta power results of reasoning type and location interaction (p < 0.05). (b) The grand average figures of event‐related theta power analysis (4–7 Hz) for O locations in the time‐frequency domain during DR (top) and PR (bottom). The red color indicates high power, and the blue color indicates low power. Asterisks indicate significant results.