On the Plurality of Thought: Beyond Dual-System Views

Abstract

The contrast between two types of thinking, one fast and automatic, the other slow and deliberate, is present in many philosophical and psychological accounts of the mind. It is also present in the common sense distinction between reason and intuition. Recent psychological data has provided support for this duality of thought. Dual-process views have been proposed for memory, reasoning, judgment, and decision-making. Researchers have sought to unify these views under a general dual-system theory: one system underlies the fast, automatic, low-effort, high-capacity processes; the other system underlies the slow, deliberate, high-effort, low-capacity processes.

In the dissertation, I show that there is no non-ad hoc way of keeping the distinction between the two systems. First, I consider the relation between systems and processes, and their role in modeling the cognitive architecture of the mind. I claim that we need to abandon the idea of a single, unified dual-system view and distinguish a duality at the micro-architectural level (a duality of processes) from a duality at the macro-architectural level (a duality of systems or duality of the mind). I show that a distinction at the micro-architectural level can explain (to a certain degree) the clustering of properties described in the dual-process literature. However, it cannot support a general dual-system view, that is, a macro-architectural duality of the mind. Appeals to underlying cognitive architecture, the use of working memory, and levels of analysis are all useful to explain some of the data and make sense of the complexity of processes involved in "higher-level" cognition. However, none of them can provide a robust account of the dual-system view. Instead, I propose a hybrid model of the mind, one that takes into account micro-architectural features and includes a third type of process, which is crucial for conflict resolution and cognitive control. To conclude, I discuss an application of the dual-system view to moral judgment and decision making. I show that, unless we have independent means of establishing the reliability of the two types of processes, there is no reason to assume, on neural data alone, that slow deliberation will be more reliable than moral intuition.