Dual-process theory

Dual-process theory holds that human thinking draws on two qualitatively different types of process. Type 1 processes are fast, autonomous and make little demand on mental capacity; Type 2 processes are slower and require working memory (Thompson et al., 2018). In its most defended form, the default-interventionist account, Type 1 processes deliver a default answer, and Type 2 processing, defined by hypothetical thinking and working-memory load, may step in and override it (Evans & Stanovich, 2013). The theory explains many errors in reasoning and judgment as Type 1 outputs that went uncorrected (Thompson et al., 2018). Two disputes surround it. One concerns sequence: correct logical answers often come first, before deliberation, which challenges the idea that logic needs Type 2 correction (Bago & De Neys, 2017). The other concerns whether two types exist at all: Melnikoff and Bargh call the typology a myth (Melnikoff & Bargh, 2018b), and leading dual-process theorists reply that their critique misses the theory (Pennycook et al., 2018).

History

Not covered by the research: the origins of the theory and how the Type 1/Type 2 terminology replaced "System 1/System 2".

Core concepts

Two types of process

Dual-process theorists distinguish processes by type, not by brain system. Type 1 processes are autonomous, run quickly and need little capacity. Type 2 processes need working memory and run more slowly. Biases are attributed to Type 1 outputs, which Type 2 processing may or may not correct (Thompson et al., 2018). Evans and Stanovich define Type 2 processing by two features, hypothetical thinking (reasoning about possibilities not given in the situation) and load on working memory (Evans & Stanovich, 2013).

Default-interventionist and parallel models

In the default-interventionist model, Type 1 gives a default response first and Type 2 may intervene afterwards (Evans & Stanovich, 2013). Accounts differ on whether a logical response can itself be produced by Type 1, as an “intuitive logic”, or only by reflective Type 2 thinking (Kosourikhina & Handley, 2025).

Applications

The same distinction is used to explain belief bias in reasoning, heuristic judgments in judgment and decision-making, and insight in problem solving, where whether insight is a distinct, unconscious Type 1 process is debated (Stuyck et al., 2022).

Questions, controversies, and new developments

Does intuition come first?

The default-interventionist account predicts that correct logical answers, when they conflict with intuition, should arrive late, after Type 2 has overridden the default. Bago and De Neys found that correct logical responses are often given as the first, immediate response (Bago & De Neys, 2017). Thompson and colleagues found that high-capacity reasoners favoured logic over belief in conflict problems even under a deadline, and conclude that part of the link between capacity and accuracy lies in better intuitions, not more Type 2 engagement (Thompson et al., 2018). A 2025 study found that conflict in reasoning arises between two heuristic responses rather than between logic and a heuristic (Kosourikhina & Handley, 2025). Whether reasoners who give biased answers still detect a conflict with logic is disputed: Mata and Ferreira argue that detection should not be expected in those who miss the critical premises, while leaving implicit detection open (Mata & Ferreira, 2018).

Do the features cluster into two types?

Melnikoff and Bargh argue that the dual-process typology is an unsupported myth that hinders progress. Their central argument is that dual-process theories assume the defining features of each type are aligned, but that there is no evidence these features cluster into two types, and the alignment has never been tested (Melnikoff & Bargh, 2018b).

Pennycook, De Neys, Evans, Stanovich and Thompson reply that feature alignment is an assumption contemporary dual-process theorists have already refuted, and that Melnikoff and Bargh attack a class of theories rather than any specific one (Pennycook et al., 2018). They argue that the theory does not need aligned features: it could rest on a single dichotomy (Pennycook et al., 2018). Melnikoff and Bargh rejoin that most dual-process advocates do assume feature alignment, and that a theory defined by a single feature uncorrelated with the others could make no predictions (Melnikoff & Bargh, 2018a).

Evans and Stanovich had earlier responded to five common criticisms of dual-process theory, in the same paper that defines Type 2 by hypothetical thinking and working-memory load (Evans & Stanovich, 2013). The question of whether the theory describes two natural kinds of process is contested.

Replication record

Not covered by the research: whether the core dual-process findings (conflict detection, early correct responses) have independent or multi-lab replications.

Broader connections

Dual-process theory is one of the theories that cut across several processes, alongside connectionism and Bayesian models of cognition; Architecture of the mind compares such accounts. Its Type 2 processes are defined by the limited capacity of working memory (Evans & Stanovich, 2013).

Further reading

• Evans and Stanovich (2013): the main defence of the theory, with replies to five criticisms.
• Melnikoff and Bargh (2018b), Pennycook et al. (2018) and Melnikoff and Bargh (2018a): the exchange over whether two types exist.
• Bago and De Neys (2017): the evidence that logic can come first.