Dual Process Models

Question 1

Precursors to modern multi-modal theories of human reasoning

Answer 1

1. Psychodynamic Theories - Freud (primary and secondary)
2. Associative and Productive (WIlliam James)
3. Automaticity (Bargh)

Question 2

Precursor to modern multi-modal theories of reasoning -

Psychodynamic theories

Answer 2

• Primary Process
  
  • Guided by what was important to the individual
  • Tended to be self-reinforcing
  • Sought pleasure and avoided pain
  • Was instinct based
  • Manifested in such activity as fantasy and imagination
• Secondary Process
  
  • Governed by the reality principle
  • Aimed to delay gratification or negotiate its satisfaction through the world of obstacles
  • Purposive and its requirements were culturally derived

Question 3

Precursors to modern multi-modal theories -

associative and productive theories (WIlliam James)

Answer 3

• Works in terms of images which cascade in patterns generated by associations
• Productive thought which is capable of dealing with novel situations

Question 4

Precursors to modern multi-modal theories of human reasoning -

Automaticity (Bargh)

Answer 4

• Social psychology has since the mid ‘70s begun to realise how little of human action is under conscious control
• Following Langer (1978) Bargh has questioned a view that saw human systematically and consciously processing incoming information
• He asserted that for much of our mental life to occur at all it must be automated and non-conscious
• He postulated three ‘control’ mechanisms
  
  • An automatic effect of perception on action
  • Automatic goal pursuit
  • Continual automatic evaluation of one’s experience

Question 5

Dual Processes of Learning - Reber

Answer 5

• The cognitive unconscious is a powerful information gathering system with rich epistemological functions to play in the life of an organism that possesses it.
• The neurological structures that sub-serve the implicit cognitive system(s) are evolutionarily old and preceded the explicit and conscious system(s) by a considerable amount of time.
• A gentle form of recapitulation holds in that the implicit system(s) also emerge earlier in life than the explicit
• Dissociation between the implicit and the explicit cognitive processes follow phylogenetic ontogenetic lines as displayed by:
  
  • Differing degrees of resistance to disruption of function by diseases and disorders
  • Differential degrees of individual-to-individual variation in the population
  • Differential sensitivity to factors such as age and IQ
  • Commonality of function across phyla

Question 6

Evans and Over Model (1996)

Answer 6

• The research stream that was the basis of this model is from the deductive logic paradigm
• The major publication of the first bloom of the theory was in a book titled “Rationality and Reasoning”
• Since the publication of this work Evans and Over have continued their research in a number of directions but within the deduction paradigm
• At a conference in 2006 Evans asserted that the theory will need to address multiple processes and not merely two
• The version that will be presented here is based on the 1996 formulation
• Evans notes that there exists an apparent paradox of rationality (p141). One the one hand people are very successful in achieving personal goals but on the other exhibit many errors and biases in reasoning
• Evans presents a terminology which captures the emphasis that humans are rational in both types of reasoning.
  
  • Rationality_1­= ability to achieve personal goals
  • Rationality_2­= a more limited ability to create systems of logic
• The model also asserts that the same problems are encountered whether dealing with reasoning or decision making. This joins what until this time had been two separate research endeavours.
• The model assumes a typical set of general stages:
  
  • Problem representation
  • Judgement or inference
  • Decision/action
• Evans et al assert that the most important distinction for them is between explicit and tacit cognitive processes
• Their model asserts that “tacit and parallel processes of thought combine with explicit and sequential processes in determining our actions” (1996, p143)
• The implicit system has the great advantage of being able to process vast amounts of information and allow for complex acts very rapidly
• This model essentially postulates a dual process in the human mind. That these sets of cognitive functions are active and adaptive in the same brain. This is the result of attempting to account for the research findings rather than a model driven by conceptual neatness
• Evans et al assert that the two systems are interactive. That is “conscious thinking is always shaped, directed and limited by tacit, pre-attentive processes” (1996, p146)
• Evans et al assume that the tacit processes are biologically constrained and that humans have learned what they need to do to survive in their daily lives via this process
• It is very important to realise that Rationality₁ is not under conscious control. Whereas Rationality₂ is under some degree of control but not fully under conscious control
• Evans et al draw on the research up to that date on implicit learning as a basis for their position
• Several implicit learning theorists (e.g. Reber, 1993) postulate separate learning systems viz the implicit and explicit. The implicit learning system does not register in our consciousness whereas the explicit does.
• Evans et al link these processes to the concept of intuition. They define this as “the experience of processes whose end-product only is posted in consciousness” (1996, p150). Much expert knowledge is of this type where the end product is reportable but not the vast set of steps to reach it.
• Evans et al offer tentative accounts for the representativeness and availability heuristics based on their twin rationality model. They note that to make these heuristics conscious would be very unusual thinking indeed e.g. “I can think of several examples of this event so it must happen often”(1996, p151). This is not an economical use of our limited cognitive resources.
• Evans et al note that Rationality₂ systems can learn to act in line with formal systems of logic or probability among others.
• Conscious thought is by contrast slow, of limited capacity and serial in nature. The question is to what adaptive advantage do these features point?
• Rationality₂ systems however allow humans to solve novel problems and to develop methods to address issues in a shorter timeframe that it would take for a solution to evolve!
• Rationality₂ systems bestow on humans both flexibility and foresight Both of these traits are essential for social cognition and group functioning (recall the cognitive demands of living in primate groups).
• Evans et al note that the “acquisition of effective explicit thinking skills is also very hard work, in contrast with the automatic and apparently effortless acquisition of our tacit and intuitive processes.” (1996, p154).
• The role of language in thinking is raised by Evans et al. In agreement with Reber (1993) Evans et al assert that possession of language is a pre-requisite for the human form of reflective consciousness and hence for the human faculty for explicit, rational₂ thought.
• A further point about language is its use in rationalisation. Folk psychology would assert that humans make decisions for reasons that they are conscious of and so can verbally report. But in agreement with Nisbett and Wilson (1977) Evans et al assert that we in fact have minimal access to reasoning of the rationality₁ system but must socially account for its products using the rationality₂ system.

Question 7

Sloman Model

Answer 7

• Starts from the H&B position that there are two clusters of data in the reasoning and judgement research
  
  • People are prone to a series of biases BUT
  • People also show the ability to form coherent and justifiable beliefs and carry out correct actions
• Sloman postulates two forms of computation with some acknowledgement of William James
  
  • Associative system – based on similarity and temporal structure. It acts on clusters of perceptions based on [quasi]-statistical regularities
  • Rule-based systems – computations reflect rule structure. They are capable of unbounded combinations of propositions
• Sloman recognises the importance of language in rule based systems and notes the role of syntax and semantics
• In agreement with Evans and Over, Sloman asserts the view that the associative system is focused on “achieving personal goals and the rule-based system is good at ensuring one’s conclusions are sanctified by normative theory” (2002, 382)
• Sloman provides a very interesting guideline for assessing the origin of an inference that comes into a person’s awareness
  
  • If we are aware of ONLY the result of the computation – Associative System
  • If we are aware of BOTH the result and the computation that derives it – Rule-based
• Sloman asserts that one of the most compelling pieces of evidence for two forms of reasoning. He terms this “criterion S” by which he means that “people believe two contradictory responses simultaneously” (2002, p384)
  
  • An example is the statement “technically a whale is a mammal”. Formally a whale is classified as a mammal but informally it is associated with fish
  • Sloman takes this approach to the Linda Problem. It is easy to imagine Linda as a feminist as the description resonates with the associations (of the time) with a feminist but adding the bank teller is “logical” but violates statistical rules
  • On a related matter Sloman notes that judgements of probability show a similar dual pattern of numeric values compared to non-numeric statements (e.g. 70% vs highly likely).
  • See the book by Philip Tetlock “Expert Political Judgement: How good is it? How can we know? (2005) for a very comprehensive analysis of forecasting from a psychological point of view.
• Sloman also draws on social cognition literature to demonstrate the existence of two systems of reasoning. E.g. he mentions impression formation and attitude change and refers to the separate research stream in this area that has been approaching a similar conclusion to his.
• Sloman explores the relationship between the two systems via his concepts of
  
  • Associative intrusion
  • Rule-based suppression
• Sloman notes that the rule-based system can suppress the associative system but that the associative system can and does intrude its opinion
• The reason for this situation lies in the nature of the operation of the two systems. The associative system is very fast compared to the rule-based system. This speed results in the results of the systems of operation appearing in our awareness much faster that the slower linearly sequential results of the rule-based system.
• In order for the rule-based system to exercise control over the associative system the person must be highly self-aware and committed to exercising restraint when considering the issue of the moment.
• Also Sloman notes that there are certain psychological states that facilitate rule-based deliberations. He quotes work from Epstein et al (Sloman, 2002, p391) which points out that when individuals are in a demand characteristic to be rational and so justify their conclusions subjects do in fact make more objective statements
• The rapid non-conscious response of the associative system raises an important issue. To what extent is our reasoning under our conscious control. Sloman notes that the associative system is characterised by two important features of other automated perceptual systems
  
  • It is highly encapsulated i.e. it does not tie up general processing resources
  • Regardless of effort the system inputs produce output i.e. the system once triggered goes through the process
• The empirical predictions associate with this model are very important for the testing and development of the theory. Sloman mentions that cognitive load should have greater impact on the Rule-Based system compared to the Associative System. We also note that working memory is more critical to rule-based reasoning than the associative system.
• Sloman makes it very clear that his model is not to be confounded with the distinction between deduction and induction. These are two forms of formalised reasoning and Sloman notes that the two systems he is describing impact on the premises and structure of the arguments for both induction and deduction.
• The question of why humans would need or indeed have two systems is of course a rather important one. Sloman postulates that the systems are complementary and so are not antagonistic. The associative system draws on the ‘statisical regularities’ of the world and so enables the relevant abstractions from experience which allows an efficient interaction with the world.
• The role of pleasure and pain in mentioned by Sloman when he notes that Freud also postulated two systems. Sloman states that gratification and avoidance of pain are one system and negotiating the world of obstacles is also part of human existence.

Question 8

Epstein (CEST) Model

Answer 8

• Based on Cognitive-Experiential-Self-Theory (CEST) which asserts that people adapt to their environment by means of two information processing systems;
• CEST asserts that humans:
  
  • Strive to maintain
    
    • A favourable level of self-esteem
    • Relatedness
    • Stable and coherent self-system
  • Seek pleasure and avoid pain
• Broadly the two systems are:
  
  • A preconscious experiential system and
  • A primarily conscious rational system
• CEST asserts that there is an age related shift in functioning from experiential to rational
• The experiential is said to be “older” than the rational system
• The features of the experiential system are:
  
  • Automatic
  • Responds to the concrete
  • Rapid
  • Non-verbal
  • Information processing is emotion driven
  • Highly compelling “conclusions”
  • Motivates action
  • Pre or un-conscious
  • Context specific thinking
• The experiential System is:
  
  • The major determinant of behaviour
  • Populated by schemas generalised from experience
  • The basis of intuitive thought
  • Hard to change schemas due to slow process of formation
• The Rational System is said to be “newer” than the experiential system
• The features of the Rational System are:
  
  • Deliberate and intentional
  • Deals more with abstract and abstracted issues via logical connections
  • Slow and oriented towards delaying action
  • Verbal
  • Information processing is more constrained by explicit beliefs
  • Conscious
• The rational system is:
  
  • Cognitive load sensitive
  • Populated by schemas and process derived from “cultural heritage”
  • Responsive to the need for justification via logic and evidence
  • Easier to change than the experiential – open to conscious input from evidence.
• The Experiential system underpins a person’s “world view” at a pre-conscious level. Epstein asserts that our world views contain four intuitive assessments of:
  
  • The world as benign vs. malevolent
  • The world as meaningful vs chaotic
  • Relationships are supportive vs threatening
  • The self is worthy vs. unworthy
• The features of the experiential system is the major determinant of behaviour due to its rapid processing which is experienced as less effortful and hence more efficient. Couple this rapidity with emotion alignment (it feels right) and as a species humans are predisposed to an automated experiential response in familiar situations
• Epstein (et al 1999, p465-5) reports that across a series of studies that participants were easily able to distinguish between the logical perspective (‘knew better’) and what most people would do (including them).
• He also reports the ratio bias effect and notes that absolute numbers are more ‘convincing’ than statistical probabilities.
• Epstein and his colleagues have developed a number of psychometric instruments to measure the two systems of thinking:
  
  • Constructive Thinking Inventory (CTI) which measures the frequency and adaptiveness of automatic thoughts and viewing events.
  • The Rational-Experiential Inventory (REI) which is a combination of a modified version of the Need for Cognition (NFC) scale combined with the Faith in Intuition (FI) scale. The scale is designed to measure the degree of processing in particular mode. The most recent version of the REI consists of four sub-scales (each of ten items i.e. 40 in total)
    
    • Experiential Engagement EE
    • Experiential Ability EA
    • Rational Engagement RE
    • Rational Ability RA
  • Epstein has also in line with his psychometric efforts examined the relationship of Dual Process Models and its relation to IQ. He asserts that abstract or intellectual intelligence corresponds to the rational system and that practical intelligence to the experiential system
  • The question of the interaction of the two systems is a very important one as Epstein acknowledges (1999,p474). How does the Experiential influence the rational and vice versa? Epstein has found that the experiential system has an automatic impact while the reverse is much more inconsistent
  • Epstein takes the view that both systems developed as adaptive to their particular environments in which they arose, though he is somewhat unclear on what those environments were.
  • He does point out that sensitive periods particularly in emotion development may have disproportionate impact on the experiential system.
  • Epstein’s theory is a very comprehensive one which attempts to integrate a vast amount of research. However, he does not address the Reasoning research of Evans or others.

Question 9

Stanovich and West

Answer 9

• “The Robot’s Rebellion” 2004
• Stanovich’s model is clearly and explicitly an evolutionary one. His title is designed to draw the reader’s attention to the problem of all genetically based life – the external body is merely a carrier for the “selfish gene”
• Stanovich asserts that in the case of humans the “robot” (read gene carrier) is capable of rebellion and that it is rationality which creates conditions for that rebellion.
• The beginning of Stanovich’s argument is the recognition that we in the west are only just beginning to enter the world of Darwin. Certain beliefs such as human autonomy and sufficiency are virtually wiped out by an in depth understanding of the implications of Darwin’s theory.
• We are replicators (molecules that make copies of themselves) procreation vehicles or robots. Only our reflective higher order reasoning capability offers us a way to transcend this predicament.
• Stanovich observes that given the contingent nature of human existence some other organism could have ended up dominating the world.
• Stanovich disagrees with many evolutionary psychologists who conflate the interests of the genes and the interests of humans. He would assert that humans have personal/species goals that are not identical with our genes.
• Humans are the first vehicle are far as we know which has recognised that the replicator will sacrifice the vehicle in order to replicate.
• The key concept derived from the replicator view is that we as a species are at war with ourselves. Stanovich describes this as two minds with one brain.
• Stanovich assembles a very impressive list of researchers who have, in one form or another, put forward dualistic models of reasoning/thinking/judgement.

Stanovich’s review of the published material and his own research leads him to an important distinction. What others have termed System1/Rationality₁/Experiential Stanovich terms the TASS system. TASS stands for The Autonomous Set of Systems. This emphases the twin facts that this system is in actuality a set of cognitive operations and functions and that one of their most prominent features is that they are automatic i.e. not under volitional control!

Question 10

• The TASS system: ‘The parts of the brain that ignore you” (2004, p37)

Answer 10

• The components of the TASS system are designed for rapid non-conscious response. In the case of many decisions thinking about them means that you die.
• These modules respond to their own triggering stimuli. Stanovich acknowledges the properties listed in the table but for his model he emphasises their autonomy. It is this property that makes them so problematic for the vehicle (ie us)
• Stanovich also emphasises that the TASS systems’ execution of their functions is not dependent upon input from the Analytic System
• The outputs of the TASS systems can and often are in conflict with the results of the simultaneous outputs of the analytic system. This potential for conflict is a major issue within his theory.
• Stanovich notes the work of Fodor on modularity and identifies the mandatory nature of these processes along with domain specificity as important convergences with his theory.
• Stanovich is not asserting that all TASS components are the result of evolutionary processes only that as a result of practice that they become automated.

Question 11

The Analytic System: “the little person in the head problem”

Answer 11

• The operation of the analytic system is basically characterised as the opposite of the TASS system. It is slow, serial, impacted by cognitive load problems and especially language based.
• The homonucleus problem is the problem that Freud and many others have had to confront. For example who is the coordinator of the executive function? We need an alternative way of explaining the coordinated processes so evident in the analytic system.
• Many including Pinker and Stanovich would assert that executive control is a series of if-then rules which move control to the loudest, fastest or strongest agent.
• The role of language is very important to Stanovich. He agrees with many cognitive theorists who note that the analytic system is “uniquely responsive to linguistic input” (2004, p48). Stanovich as a result notes that the serial nature of language may have been the origin of the analytic system.
• One of the key capacities of the analytic system is to represent possible future worlds and so tentatively explore varieties of optimal actions. Hypothetical thinking is a key issue here. Hypotheses are then able to be tested. Human capacity for exploring future worlds is a critical element in our adaptive-ness.

Question 12

Stanovich and West (ctd)

Answer 12

• Stanovich asserts that modern existence increasingly requires the activation of “the evaluative and supervisory functions of the analytic system to overcome the habitual responses (of the TASS system) that are no longer serving our needs” (2004, p60)
• This is the situations where minds are in conflict. The analytic system is not superior but more adaptive for many of the activities of modern life. Its task is to damp down the automated responses of the TASS system.
• It is thus the TASS system that is most supportive and facilitative of the gene goals as most of the TASS modules are of older evolutionary origin.
• Like Evans, Stanovich acknowledges the contributions of Reber.
• The analytic system is focused on the organisms needs and goals. Its functions are designed to help achieve the personal goals of the individual. This is potentially going to mean the sacrificing genetic fitness.
• Referring to the Sphex wasp Stanovich asserts that humans in fact are a little bit sphexish i.e. lots of activity that is unreflectively undertaken and which is not focused on the individual’s best interests.
• As with other theorists in this area Stanovich recognises that humans do not have two heads or existences. One physical brain is the ‘site’ of the two systems. How can this be? In fact how did System 2 or Analytic reasoning come into being?
• Stanovich is not completely clear on the origin of System 2 processes but does (2004, p48) indicate that it is a relatively recent phenomena and is heavily connected to language. One can therefore speculate that the emergence of language may have been the reason for analytical reasoning developing.
• As humans developed both culturally and cognitively the Analytic system became more and more prominent. Yet this development did not result in the diminishment of TASS. Humans thus have two competing minds.
• TASS is good at;
  
  • Positive instances of events and relationships
  • Highly practiced
  • Frequency coded
  • Time pressured
  • Recognition based
• TASS is responsible for (Stanovich, 2004, p113);
  
  • “the tendency to contextualise a problem with as much prior knowledge as is easily accessible, even when the problem is formal and the only solution is a content-free rule,
  • The tendency to socialise problems even in situations where interpersonal cues are few,
  • The tendency to see deliberate design and pattern in situations that lack intentional design and pattern, and
  • The tendency toward a narrative mode of thought”

Question 13

Memes: Beliefs that replicate

Answer 13

• Memes are essentially instructions for behaviours that can be learned by imitation and stored in our brains
• While genes hold the instructions for building our bodies that carry the genes, memes hold the instructions for building the culture that the bodies exist in and hence facilitate replication
• The question is then how many of our beliefs are junk beliefs (reference to junk DNA which services no known purpose).
• Memes have the following features;
  
  • Most memes reflect true information therefore that are helpful
  • Certain memes fit well with evolutionary modules
  • Certain memes spread because they facilitate replication of genes
  • Memes have self-replicating properties
• Some memes however set up expectations that are anathema to reflection. They are about simple reflexion i.e. non-contemplative in their enactment
• For progress to be made in the utilisation of Analytic processes people must reject memes that obstruct reflection and processes that rigorously examine beliefs and experience.
• Stereotypes and rigid worldviews are memes.

Question 14

Wastell’s Multi-Modular Set Theory

Answer 14

• Wastell (in press) has proposed a different solution to the control question in reasoning. He builds on modular approaches such as Stanovich;s but combines it with emergence theory to deal with the homonucleus problem.
• Stanovich restricts modules to System 1 or the TASS. There is no logical or empirical reason for such a restriction. In other words higher order reasoning can be potentially viewed as module in function.
• This view is consistent with those theorists who propose a massively modular view and so asserts that human reasoning is potentially a very fragemented process unless there is some form of coordination.
• Human reasoning does not normally show features that one would describe as fragmented. How is this possible if the mind consists of a large number of informationally encapsulated modules?
• What is needed is a system of coordination that is produced by simple rules that are contained within the modules themselves
• This is where “emergence” theory provides a viable alternative to executive control approaches.
• Emergence theory is based on the proposition that complex behaviour can emerge from a combination of simple rules
• These rules are internal to the unit of behaviour. This means that in aggregrate the rules can be quite efficient as there is no need for nor mechanism of centralised control which is both time consuming and requires knowledge of the entire system in order to adequately exercise control.
• An example of emergence theory is the macro behaviour of a flock of birds avoiding a predator. Each bird obey four simple rules and there is no central control. The four rules:
  
  • Avoid the predator
  • Avoid the ground
  • Move toward the centre of the flock
  • Stay safe flying distance from all their neighbours
• These four rules result in the displays of what appears to be coordinated behaviour as the flock avoids the predator. It is also the same for a school of fish.
• The application of emergence theory to human reasoning as suggested by Wastell (in press) is at a very early stage. Wastell designates these as Virtual Modules to distinguish them from innate modules proposed by authors such as Fodor (1983), Carruthers (2003) and Coltheart (1999).
• Wastell suggests that an important aspect of reasoning modules is their information requirements. Innate modules are “informationally encapsulated” meaning that they are triggered by incoming information [e.g. face recognition characteristics].
• Wastell’s model asserts that the information output of one module becomes the input for another module. The combination of a number of reasoning sub-components performing as virtual modules would then “appear” as a single process.
• There are many theoretical and empirical questions to be answered. Some of the most pressing questions are:
  
  • What is the structure of higher order reasoning modules?
  • To what extent are they learnt?
  • What are the mechanisms that lead to the formation of a reasoning module?
  • How complex can a module become it must divide into simpler sub-components?
• Wastell’s model requires empirical studies to ascertain its contribution to the field but it does propose a new multi-theory approach to the study of human reasoning.