Rethinking Dual Process Models of Thought

This is the fourth in a series of posts on my book, The Four Realms of Existence. Read posts one, two, and three.

Source: Harvard University Press/Used with permission

Everything about a human being, biologically and psychologically, can be subsumed within four fundamental realms of existence. These are the biological, neurobiological, cognitive, and conscious realms. In writing about the cognitive realm, I came to question aspects of the popular distinction between two kinds of processes underlying thought, labeled in various ways: fast versus slow, automatic versus effortful, implicit versus explicit, intuitive versus deliberate, and conscious versus non-conscious. In particular, I believe that the various processes might be fruitfully viewed as being distributed across three systems.

There are many dual-process models of thought, but psychologist Daniel Kahneman’s two-system perspective has been very influential in recent years and is useful for introducing the topic.

According to Kahneman, System 1 is fast and includes innate capacities we’ve inherited from, and share with, other mammals. He says, “We are born ready to perceive the world around us, recognize objects, orient attention, avoid losses, and fear spiders.” But his System 1 is not purely innate. It also learns associations associated with emotional behavior. His System 2, by contrast, is slow and uses the deliberative powers of working memory to explore alternative decisions and actions. It allocates attention to activities that require mental work, and its functions are closely associated with agency, choice, concentration, and conscious experience. The upside of fast System 1 processes is that they are computationally and energetically cheap. The upside of System 2 processes is that they make possible purposive deliberation to consciously pursue goals, but they come with high computational and energy costs.

An earlier dual-process perspective comes from psychologist Jonathan Evans and colleagues. It also proposes two processes, which, for simplicity, I will call System 1 and 2. For Evans, System 1 is “implicit,” “associative,” “early evolved,” “old brain,” and “similar to animal cognition”. System 2, by contrast, depends on working memory. As Evans pointed out, System 2 processes are a natural psychological category because they all require working memory. System 1 processes, on the other hand, have no obvious underlying psychological unifying factor; they are a grab-bag collection of processes, the only common feature of which is that they do not depend on working memory.

On the surface, this seems straightforward. But if we dig deeper, there is, in fact, a natural category within System 1 that is different from other System 1 processes. This is the category Evans and Kahneman both described as cognition inherited from our animal ancestors. Given their emphasis on working memory and cognition, they must be assuming mammals lack working memory. But monkeys have non-verbal working memory capacities quite similar to humans, and even non-primate mammals have primitive working memory capacities. It would seem that all processes dependent on working memory, even if it is a primitive kind of animal cognition in the human brain, should be in System 2.

Much has been learned about the neural basis of working memory processes. In particular, areas of the lateral prefrontal cortex have long been known to contribute to working memory in humans and other primates. While non-primate mammals lack a lateral prefrontal cortex, their working memory capacities are mediated by the medial prefrontal cortex. In humans, areas of the medial prefrontal cortex have been implicated in various cognitive processes, like attention and conflict resolution in decision-making and action selection, that contribute to lateral prefrontal working memory functions and may account for what was described as "similar to animal cognition."

Kahneman and Evans may have been referring to non-conscious cognition when they mentioned cognition similar to animals since both authors include consciousness as a hallmark of System 2. But as Evans pointed out, some of the workings of System 2 are unconscious. This view meshes well with what John Kihlstrom referred to as the "cognitive unconscious."

Given that System 2 has both conscious and non-conscious cognitive processes, it might be useful to make non-conscious cognition the subject of System 2 and move conscious cognition to a third system. The resulting three systems, labeled with letters rather than numbers, would be:

• System A: Non-Cognitive and Non-Conscious Processes. This would include implicit non-cognitive processes that are automatically controlled.
• System B: Cognitive But Not Conscious Processes. Here, we'd find implicit cognitive processes like non-conscious intuition, non-conscious inferential reasoning, non-conscious deliberation, non-conscious working memory, and non-conscious mental models.
• System C: Cognitive and Conscious Processes. This system would encompass conscious experiences that reflect underlying working memory and mental models.

An important advantage of the three-system scheme is that it provides a framework for understanding levels of behavioral control in the human brain:

• System A: Not Cognitive and Not Conscious Behavioral Control. Includes non-conscious processes that control reflexes, innate behaviors, and habits, mostly via subcortical brain areas.
• System B: Cognitive But Not Conscious Behavioral Control. Consists of non-conscious cognitive processes that control goal-directed behavior, including speech.
• System C: Cognitive and Conscious Behavioral Control. This would encompass conscious mental states that control goal-directed behavior, including speech.

With regards to my book, what we end up with is that each system constitutes a different evolutionarily-derived realm of existence, and a corresponding set of functions, in the human brain.

• System A: The Neurobiological Realm
• System B: The Mere (Non-Conscious) Cognitive Realm
• System C: The Conscious Realm

Cognition Essential Reads

Are Large Language Models the New "Thinking Aids"?

Why Do People Fall for Vague Promises?

Some may be confused about how something can be cognitive but not conscious. In the 1950s, Karl Lashley noted that every conscious experience is preceded by non-conscious information processing. But not all non-conscious cognitive processing results in conscious experience. For example, in a typical conversation, there is some topic being discussed. Relevant memories bundled together as non-conscious schema serve as a template (mental model) for thought, speech, and action. As a result, you can converse back and forth without having to explicitly think consciously about exactly what you are saying (System B at work). But if the topic veers, or you disagree with the other person, then you probably switch to deliberate, conscious consideration of where to take things (System C kicks in). This restructures the relevant mental model, allowing you to slip back into non-conscious, non-deliberative cognitive thinking and talking (System B takes over) until you again need to deliberate (System C in charge), and on and on.

For more details on the relation of the three systems to the realms of existence in the human brain, please refer to the book.