Dual process theory

In psychology, a dual process theory provides an account of how a phenomenon can occur in two different ways, or as a result of two different processes. Often, the two processes consist of an implicit (automatic), unconscious process and an explicit (controlled), conscious process. Verbalized explicit processes or attitudes and actions may change with persuasion or education; though implicit process or attitudes usually take a long amount of time to change with the forming of new habits. Dual process theories can be found in social, personality, cognitive, and clinical psychology. It has also been linked with economics via prospect theory and behavioral economics.

History

The foundations of dual process theory likely comes from William James. He believed that there were two different kinds of thinking: associative and true reasoning. James theorized that empirical thought was used for things like art and design work. For James, images and thoughts would come to mind of past experiences, providing ideas of comparison or abstractions. He claimed that associative knowledge was only from past experiences describing it as “only reproductive”. James believed that true reasoning was useful for “unprecedented situations” in which using reasoning to overcome obstacles such as navigation could be overcome with reasoning power of being able to use a map.

There are various dual process theories that were produced after William James's work. Dual process models are very common in the study of social psychological variables, such as attitude change. Examples include Petty and Cacioppo's Elaboration Likelihood Model explained briefly below and Chaiken's Heuristic Systematic Model. According to these models, persuasion may occur after either intense scrutiny or extremely superficial thinking. In cognitive psychology, attention and working memory have also been conceptualized as relying on two distinct processes.^[1] Whether the focus be on social psychology or cognitive psychology, there are many examples of dual process theories produced throughout the past. The following just show a glimpse into the variety that can be found.

Jonathan Evans suggested dual process theory in 1984. In his theory, there are two distinct types of processes: heuristic processes and analytic processes. He suggested that during heuristic processes, an individual chooses which information is relevant to the current situation. Relevant information is then processed further whereas irrelevant information is not. Following the heuristic processes come analytic processes. During analytic processes, the relevant information that is chosen during the heuristic processes is then used to make judgments about the situation.^[2]

Richard E. Petty and John Cacioppo proposed a dual process theory focused in the field of social psychology in 1986. Their theory is called the Elaboration Likelihood Model of Persuasion. In their theory, there are two different routes to persuasion in making decisions. The first route is known as the central route and this takes place when a person is thinking carefully about a situation, elaborating on the information they are given, and creating an argument. This route occurs when an individual’s motivation and ability are high. The second route is known as the peripheral route and this takes place when a person is not thinking carefully about a situation and uses shortcuts to make judgments. This route occurs when an individual’s motivation and ability are low.^[3]

Steven Sloman produced another interpretation on dual processing in 1996. He believed that associative reasoning takes stimuli and divides it into logical clusters of information based on statistical regularity. He proposed that how you associate is directly proportional to the similarity of past experiences, relying on temporal and similarity relations to determine reasoning rather than an underlying mechanical structure. The other reasoning process in Sloman's opinion was of the Rule based system. The system functioned on logical structure and variables based upon rule systems to come to conclusions different from that of the associative system. He also believed that the Rule based system had control over the associative system, though it could only suppress it.^[4] This interpretation corresponds well to earlier work on computational models of dual processes of reasoning.^[5]

Daniel Kahneman provided further interpretation by differentiating the two styles of processing more, calling them intuition and reasoning in 2003. Intuition (or system 1), similar to associative reasoning, was determined to be fast and automatic, usually with strong emotional bonds included in the reasoning process. Kahneman said that this kind of reasoning was based on formed habits and very difficult to change or manipulate. Reasoning (or system 2) was slower and much more volatile, being subject to conscious judgments and attitudes.^[6]

Fritz Strack and Roland Deutsch proposed another dual process theory focused in the field of social psychology in 2004. According to their model, there are two separate systems: the reflective system and the impulsive system. In the reflective system, decisions are made using knowledge and the information that is coming in from the situation is processed. On the other hand in the impulsive system, decisions are made using schemes and there is little or no thought required.^[7]

Theories

Dual process learning model

Ron Sun proposed a dual process model of learning (both implicit learning and explicit learning). The model (named CLARION) re-interpreted voluminous behavioral data in psychological studies of implicit learning and skill acquisition in general. The resulting theory is two-level and interactive, based on the idea of the interaction of one-shot explicit rule learning (i.e., explicit learning) and gradual implicit tuning through reinforcement (i.e. implicit learning), and it accounts for many previously unexplained cognitive data and phenomena based on the interaction of implicit and explicit learning.^[8]

Dual coding

Using a somewhat different approach, Allan Paivio has developed a dual-coding theory of information processing. According to this model, cognition involves the coordinated activity of two independent, but connected systems, a nonverbal system and a verbal system that is specialized to deal with language. The nonverbal system is hypothesized to have developed earlier in evolution. Both systems rely on different areas of the brain. Paivio has reported evidence that nonverbal, visual images are processed more efficiently and are approximately twice as memorable. Additionally, the verbal and nonverbal systems are additive, so one can improve memory by using both types of information during learning.^[9]

Dual-process accounts of reasoning

Background

The dual-process accounts of reasoning posits that there are two systems or minds in one brain. The theory of two distinct kinds of reasoning has been around for as long as documentations about theories of reasoning go. The current theory is that there are two distinctively separate cognitive systems underlying thinking and reasoning and that these different systems were developed through evolution.^[10] These systems are often referred to as being either implicit or explicit, however some theorists prefer to emphasize the functional differences between the two systems and not the consciousness factor and thus refer to the systems simply as System 1 and System 2. The broad terms System 1 and System 2 were coined by Stanovich and West^[11] and will be used throughout this article.

Systems

The systems have multiple names by which they can be called, as well as many different properties.

Differences between System 1 and System 2^[12]

System 1	System 2
Unconscious reasoning	Conscious reasoning
Implicit	Explicit
Automatic	Controlled
Low Effort	High Effort
Large capacity	Small capacity
Rapid	Slow
Default Process	Inhibitory
Associative	Rule based
Contextualized	Abstract
Domain Specific	Domain General
Evolutionarily Old	Evolutionarily recent
Nonverbal	Linked to language
Includes recognition, perception, orientation	Includes rule following, comparisons, weighing of options
Modular Cognition	Fluid Intelligence
Independent of working memory	Limited by working memory capacity
Non-Logical	Logical
Parallel	Serial

System 1

Bargh (1994) reconceptualized the notion of an automatic process by breaking down the term “automatic” into four components: awareness, intentionality, efficiency, and controllability. One way for a process to be labeled as automatic is for the person to be unaware of it. There are three ways in which a person may be unaware of a mental process: they can be unaware of the presence of the stimulus (subliminal), how the stimulus is categorized or interpreted (unaware of the activation of stereotype or trait constructs), or the effect the stimulus has on the person’s judgments or actions (misattribution). Another way for a mental process to be labeled as automatic is for it to be unintentional. Intentionality refers to the conscious “start up” of a process. An automatic process may begin without the personal consciously willing it to start. The third component of automaticity is efficiency. Efficiency refers to the amount of cognitive resources required for a process. An automatic process is efficient because it requires few resources. The fourth component is controllability, referring to the person’s conscious ability to stop a process. An automatic process is uncontrollable, meaning that the process will run until completion and the person will not be able to stop it. Bargh (1994) conceptualizes automaticity as a component view (any combination awareness, intention, efficiency, and control) as opposed to the historical concept of automaticity as an all-or-none dichotomy.^[13]

System 2 In Humans

System 2 is evolutionarily recent and specific to humans. It is also known as the explicit system, the rule-based system, the rational system,^[10] or the analytic system.^[14] It performs the more slow and sequential thinking. It is domain-general, performed in the central working memory system. Because of this, it has a limited capacity and is slower than System 1 which correlates it with general intelligence. It is known as the rational system because it reasons according to logical standards.^[14] Some overall properties associated with System 2 are that it is rule-based, analytic, controlled, demanding of cognitive capacity, and slow.^[10]

Social psychology

The dual process has impact on social psychology in such domains as stereotyping, categorization, and judgment. Especially, the study of automaticity and of implicit in dual process theories has the most influence on a person's perception. People usually perceive other people's information and categorize them by age, gender, race, or role. According to Neuberg and Fiske (1987) a perceiver who receives a good amount of information about the target person then will use their formal mental category (Unconscious) as a basis for judging the person. When the perceiver is distracted, the perceiver has to pay more attention to target information (Conscious).^[15] Categorization is the basic process of stereotyping in which people are categorized into social groups that have specific stereotypes associated with them.^[16] It is able to retrieve people's judgment automatically without subjective intention or effort. Attitude can also be activated spontaneously by the object. John Bargh's study offered an alternative view, holding that essentially all attitudes, even weak ones are capable of automatic activation. Whether the attitude is formed automatically or operates with effort and control, it can still bias further processing of information about the object and direct the perceivers’ actions with regard to the target. According to Shelly Chaiken, heuristic processing is the activation and application of judgmental rules and heuristics are presumed to be learned and stored in memory. It is used when people are making accessible decisions such as “experts are always right” (system 1) and systematic processing is inactive when individuals make effortful scrutiny of all the relevant information which requires cognitive thinking (system 2).^[17] The heuristic and systematic processing then influence the domain of attitude change and social influence.

Stereotyping

The stereotyping dual process model proposes that when we perceive an individual the stereotypes that pertain to them are automatically activated, and these stereotypes will be used in judgment when there is no motivation or cognitive capacity to inhibit the stereotype. However, controlled cognitive processes can inhibit the use of stereotypes when there is motivation and cognitive resources to do so. Devine (1989) provided evidence for the dual process theory of stereotyping in a series of three studies. Study 1 tested for the knowledge of cultural stereotypes and found that participants who were high or low in prejudice (according to the Modern Racism Scale) were equally knowledgeable of the cultural stereotypes of African Americans. Study 2 was interested in seeing whether people would use automatically activated stereotypes in their judgments regardless of prejudice level (personal belief). Participants were primed with stereotype relevant or non-relevant words and then asked to give hostility ratings of a target with an unspecified race who was performing ambiguously hostile behaviors. Regardless of prejudice level, participants who were primed with more stereotype relevant words gave higher hostility ratings to the ambiguous target. Study 3 investigated whether people can control for stereotype use with their own personal beliefs. Participants were asked to do a thought listing of African Americans, and those participants who were low in prejudice thought with less stereotype consistent than those high in prejudice. These three studies taken together show that regardless of personal beliefs when individuals do not have the motivation or cognitive capacity to inhibit stereotype use then they will go on to use the stereotype in judgments of others, thereby using automatic processes in decision making. However, when people have the cognitive resources and motivation to inhibit stereotypes by using personal beliefs then the stereotype use will be inhibited.^[18]

Decision Making

Dijksterhuis and Nordgren (2006) presented a dual process model of decision making. Based on their own and others’ previous research, Dijksterhuis and Nordgren formed the Unconscious thought theory (UTT). UTT can be broken down into six principles that relate to unconscious and conscious thought: the unconscious thought principle, the capacity principle, the bottom-up-versus-top-down principle, the weighting principle, the rule principle, and the convergence-versus-divergence principle. Further details on the six principles can be found on the Unconscious thought theory page. ^[19]

The 6 principles in relation to system 1 and system 2

System 1 relates to unconscious thought (principle 1), has a greater capacity as described (principle 2), is a more bottom up process (principle 3), is more likely to assign appropriate weights to decisions criteria (principle 4), is unable to follow one certain rule (principle 5), and is more divergent (principle 6).

System 2 can be described as conscious thought (principle 1), has a smaller capacity (principle 2), has a top-down focus (principle 3), places inappropriate weights on decision criteria (principle 4), is able to follow one specific rule (principle 5), and is more focused and convergent (principle 6).

The Relationship between Unconscious Thought Theory and Decision Making

Satisficing (SAT)

Dijksterhuis and Nordgren (2006) explain satisficing as comparing choice alternatives to a predetermined standard and if the choice alternatives do not meet the standard they are dismissed. If a choice alternative does meet the standards for all attributes then that alternative is chosen. It is difficult to map this strategy onto conscious and unconscious thought. For this strategy, it is a requirement that a single option is considered at a time. This rule is very strict and can only be accomplished by conscious thought and not unconscious thought. SAT also requires accuracy in determining whether or not something meets a predetermined standard. For SAT to be successful the right standard has to be set. Conscious thought is capable of this as long as the standard is easy to determine and easy to put into words. For example, when someone is shopping for a car, the first car that meets all of their requirements is the car that will be chosen: additional options will be unnecessary and therefore not pursued.

The lexicographic strategy (LEX)

As explained in Dijksterhuis and Nordgren (2006) in this strategy, people select their choice with the best value on the most important dimension. If the size of your new house matters a great deal more to you than all other features, choose the largest house. This strategy follows one strict rule, making conscious thought rather than unconscious thought the preferred mode of thinking. There is one catch in that consciousness has to be able to correctly identify the most important dimension. But when that requirement is met, conscious thought is more suitable than unconscious thought for LEX.

The weighted adding strategy (WADD)

Dijksterhuis and Nordgren (2006) state that “the weighted adding strategy (WADD) is a complex strategy in which a chooser first assesses the importance of different attributes, then assigns a choice alternative a value on each attribute, and then multiplies each value for that alternative by the importance of the attribute dimension. The resulting scores for all attributes are summed to obtain a score for the attractiveness of the choice alternative, and the process is repeated for the other alternatives.” WADD is very complex and therefore most useful for complex problems. Furthermore, since WADD is complex, using it unconsciously is preferable according to the capacity principle because conscious thought cannot deal with the large amount of information necessary. An example of WADD is choosing a vacation destination and weighing the pros and cons for each destination to obtain the most attractive vacation destination.

Summary of UTT

An important aspect of this theory is that conscious thought is needed when gathering all the information and then the information can be coded either consciously or unconsciously, a decision that should be made based on the complexity of the issue at hand. They propose that conscious thought is more efficient at dealing with simple issues, whereas unconscious thought is better at dealing with complex matters. Complex issues are best dealt with unconsciously because of the holistic approach, the higher capacity for processing the information, and because no weights are assigned to any particular attributes. Conversely, simple issues are best dealt with consciously because of the focused and convergent attention and the ability to actively follow rules

Evidence

Belief Bias Effect

A belief bias is the tendency to judge the strength of arguments based on the plausibility of their conclusion rather than how strongly they support that conclusion.^[20] There is some evidence that suggests this bias results from the competition between a logical process (System 2) and a belief-based process (System 1) when evaluating the argument.

Studies on belief-bias effect were first designed by Jonathan Evans to create a conflict between logical reasoning and prior knowledge about the truth of conclusions. In studies, participants are asked to assess the logical validity of arguments, i.e. whether the conclusions can be drawn from a given set of premises. The critical manipulations are that 1) the argument is either valid or invalid, and 2) the conclusions are either believable or unbelievable. Participants are asked to evaluate syllogisms that are: valid arguments with believable conclusions, valid arguments with unbelievable conclusions, invalid arguments with believable conclusions, and invalid arguments with unbelievable conclusions.^[10] Participants are told to only agree with conclusions that logically follow from the premises given. The results suggest when there is an invalid argument but the conclusion is believable, people erroneously accept the conclusion as valid even though it does not logically follow from the premises. This is belief-bias, and suggest prior knowledge from System 1 interferes with the logic of System 2.^[10]

Tests with Working Memory

De Neys^[21] conducted a study that manipulated working memory capacity while answering syllogistic problems. This was done by burdening executive processes with secondary tasks. Results showed that when System 1 triggered the correct response, the distractor task had no effect on the production of a correct answer which supports the fact that System 1 is automatic and works independently of working memory, but when belief-bias was present (System 1 belief-based response was different from the logically correct System 2 response) the participants performance was impeded by the decreased availability of working memory. This falls in accordance with the knowledge about System 1 and System 2 of the dual-process accounts of reasoning because System 1 was shown to work independent of working memory, and System 2 was impeded due to a lack of working memory space so System 1 took over which resulted in a belief-bias.^[21]

fMRI Studies

Vinod Goel produced neuropsychological evidence for dual-process accounts of reasoning using fMRI^[22] studies. They provided evidence that anatomically distinct parts of the brain were responsible for the two different kinds of reasoning. They found that content-based reasoning caused left temporal hemisphere activation whereas abstract formal problem reasoning activated the parietal system. They concluded that different kinds of reasoning, depending on the semantic content, activated one of two different systems in the brain.^[22]
A similar study incorporated fMRI during a belief-bias test.^[23] They found that different mental processes were competing for control of the response to the problems given in the belief-bias test. The prefrontal cortex was critical in detecting and resolving conflicts, which are characteristic of System 2, and had already been associated with that System 2. The ventral medial prefrontal cortex, known to be associated with the more intuitive or heuristic responses of System 1, was the area in competition with the prefrontal cortex.^[23]

Near-infrared Spectroscopy

Tsujii and Watanabe^[14] did a follow-up study to Goel and Dolan's^[23] fMRI experiment. They examined the neural correlates on the inferior frontal cortex (IFC) activity in belief-bias reasoning using near-infrared spectroscopy (NIRS). Subjects performed a syllogistic reasoning task, using congruent and incongruent syllogisms while attending to an attention-demanding secondary task. Their interest was in how the secondary-tasks changed the activity of the IFC during congruent and incongruent reasoning processes. The results showed that the participants performed better in the congruent test than the incongruent test (evidence for belief-bias), the high demand secondary test impaired incongruent reasoning results but not congruent reasoning. NIRS results showed that the right IFC was activated more during incongruent trials. Participants with enhanced right IFC activity performed better on the incongruent reasoning than those with decreased right IFC activity. This study provided some evidence to enhance the fMRI results that the right IFC, specifically, is critical in resolving conflicting reasoning but is attention-demanding, its effectiveness decreases with loss of attention where the automatic heuristic System 1 takes over which results in a belief-biases.^[14]

Matching Bias

Matching bias is a non-logical heuristic.^[24] The matching bias is described as a tendency to use lexical content matching of the statement, about which one is reasoning, to be seen as relevant information and do the opposite as well, ignore relevant information that doesn't match. It mostly affects problems with abstract content. It doesn't involve prior knowledge and beliefs but it is still seen as a System 1 heuristic that competes with the logical System 2.^[24]

The Wason selection task provides evidence for the matching bias.^[10] The test is designed as a measure of a person's logical thinking ability.^[25] Performance on the Wason Selection Task is sensitive to the content and context with which it is presented. If you introduce a negative component into the conditional statement of the Wason Selection Task, e.g. 'If there is an A one side of the card then there is not a 3 on the other side', there is a strong tendency to choose cards that match the items in the negative condition to test, regardless of their logical status. Changing the test to be a test of following rules rather than truth and falsity is another condition where the participants will ignore the logic because they will simply follow the rule, e.g. changing the test to be a test of a police officer looking for underaged drinkers.^[24] The original task is more difficult because it requires explicit and abstract logical thought from System 2, and the police officer test is cued by relevant prior knowledge from System 1.^[10]
Studies have shown that you can train people to inhibit matching bias which provides neuropsychological evidence for the dual-process theory of reasoning.^[10] When you compare trials before and after the training there is evidence for a forward shift in activated brain area. Pre-test results showed activation in locations along the ventral pathway and post-test results showed activation around the ventro-medial prefrontal cortex and anterior cingulate.^[26] Matching bias has also been shown to generalise to syllogistic reasoning.^[27]

Evolution

Dual-process theorists claim that System 2, a general purpose reasoning system, evolved late and worked alongside the older autonomous sub-systems of System 1.^[28] The success of Homo Sapiens Sapiens lends evidence to their higher cognitive abilities above other hominids. Mithen theorizes that the increase in cognitive ability occurred 50,000 years ago when representational art, imagery, and the design of tools and artefacts are first documented. She hypothesizes that this change was due to the adaptation of System 2.^[28]
Most evolutionary psychologists do not agree with dual-process theorists. They claim that the mind is modular, and domain-specific, thus they disagree with the theory of the general reasoning ability of System 2. They have difficulty agreeing that there are two distinct ways of reasoning and that one is evolutionarily old, and the other is new.^[10] To ease this discomfort, the theory is that once System 2 evolved, it became a 'long leash' system without much genetic control which allowed humans to pursue their individual goals.^[11]

Issues with the Dual-Process Account of Reasoning

The dual-process account of reasoning is an old theory, as noted above. But according to Evans^[29] it has adapted itself from the old, logicist paradigm, to the new theories that apply to other kinds of reasoning as well. And the theory seems more influential now than in the past which is questionable. Evans outlined 5 "fallacies":

1. All dual-process theories are essentially the same. There is a tendency to assume all theories that propose two modes or styles of thinking are related and so they end up all lumped under the umbrella term of "dual-process theories".
2. There are just two systems underlying System 1 and System 2 processing. There are clearly more than just two cognitive systems underlying people's performance on dual-processing tasks. Hence the change to theorizing that processing is done in two minds that have different evolutionary histories and that each have multiple sub-systems.
3. System 1 processes are responsible for cognitive biases; System 2 processes are responsible for normatively correct responding. Both System 1 and System 2 processing can lead to normative answers and both can involve cognitive biases.
4. System 1 processing is contextualised while System 2 processing is abstract.^[29] Recent research has found that beliefs and context can influence System 2 processing as well as System 1.^[30]
5. Fast processing indicates the use of System 1 rather than System 2 processes. Just because a processing is fast does not mean it is done by System 1. Experience and different heuristics can influence System 2 processing to go faster.^[29]

Another argument against dual-process accounts for reasoning which was outlined by Osman is that the proposed dichotomy of System 1 and System 2 does not adequately accommodate the range of processes accomplished.^[31] Moshman proposed that there should be four possible types of processing as opposed to two. They would be implicit heuristic processing, implicit rule-based processing, explicit heuristic processing, and explicit rule-based processing.^[32]
In response to the question as to whether there are dichotomous processing types, many have instead proposed a single-system framework which incorporates a continuum between implicit and explicit processes.^[31]

Alternative Model

The dynamic graded continuum (DGC), originally proposed by Cleeremans and Jiménez is an alternative single system framework to the dual-process account of reasoning. It has not been accepted as better than the dual-process theory; it is instead usually used as a comparison with which one can evaluate the dual-process model. The DGC proposes that differences in representation generate variation in forms of reasoning without assuming a multiple system framework. It describes how graded properties of the representations that are generated while reasoning result in the different types of reasoning. It separates terms like implicit and automatic processing where the dual-process model uses the terms interchangeably to refer to the whole of System 1. Instead the DGC uses a continuum of reasoning that moves from implicit, to explicit, to automatic.^[31]

Fuzzy-Trace theory

According to Charles Brainerd and Valerie Reyna’s Fuzzy-trace theory of memory and reasoning, people have two memory representations: verbatim and gist. Verbatim is memory for surface information (e.g.-the words in this sentence) whereas gist is memory for semantic information (e.g.-the meaning of this sentence). This dual process theory posits that we encode, store, retrieve, and forget the information in these two traces of memory separately and completely independent of each other. Furthermore, the two memory traces decay at different rates: verbatim decays quickly, while gist lasts longer. In terms of reasoning, Fuzzy-Trace Theory posits that as we mature we increasingly rely more on gist information over verbatim information. Evidence for this lies in framing experiments where framing effects become stronger when verbatim information(percentages) are replaced with gist descriptions.^[33]

See also

• Automatic and Controlled Processes (ACP)
• Cognitive inhibition
• Fuzzy-trace theory
• Opponent-process theory
• Elaboration Likelihood Model

External links

• Laboratory for Rational Decision Making

References