Intuition and Dual Process Theory

 

Philosophers define intuition as rational and superior to analytical thinking primarily because it is anchored in Ideas, Forms and Archetypes, which are perceived as a priori laws governing and conditioning all existence. Psychologists disagree and tend to equate intuition with rapid, automatic, and often biased processing. That is the case in dual process theories as well. In the chapter on intuition in philosophy we also reached the tentative conclusion that our mind is either dualistic in its functioning or it is not.[1] In the latter case, it is operating on the suggested third level of intuition. This distinction is in many ways with us in modern psychology and it is intrinsic to dual process theories. The key differences in the properties of these two processes as listed by Stanowich and West are presented in the table below.[2]

 

 

Dual-Process Theories

System 1

System 2

Hammond 1996 Intuitive Cognition Analytical Cognition
  Reber 1993   Implicit Cognition  Explicit Cognition
Johnson-Laird 1983 Implicit Inferences Explicit Inferences
Evans & Over 1996 Tacit Thought Processes Explicit Thought Processes
Sloman 1996 Associative System Rule-based System
Evans 1984, 1989 Heuristic Processing Analytic Processing
Levinson 1995 Interactional Intelligence Analytic Intelligence
Epstein 1994,   1996 Intuitive-Experiential System Rational System
Pollock 1991 Quick & Inflexible Modules Intellection
Klein 1998 Recognition-Primed Decisions Rational Choice Strategy
Shiffrin & Schneider 1977 Automatic Processing Controlled Processing
Posner & Snyder 1975 Automatic Activation Conscious Processing System

 

 

System 1 then, is characterized as intuitive, holistic, largely unconscious, and relatively undemanding of computational capacity. “It conjoins properties of automatic and heuristic processing as these constructs have been variously discussed in the literature. This system has as its goal the ability to model other minds in order to read intention and to make rapid interactional moves based on those modeled intentions.” System 2 is characterized as analytical, controlled processing. “It encompasses the processes that have been studied by information processing theorists trying to uncover the computational components underlying intelligence.”[3] As opposed to system 1, it is demanding of cognitive capacity. It is a relatively slow acquisition by cultural and formal tuition. Stanovich & West argue that system 1, on the other hand, is a relatively fast acquisition by biology, exposure, and personal experience. A general concern with the intuitive component of system 1 is its rather weak theoretical platform. This is reflected in Epstein’s work, where face validity of the intuition scale is the sole means of validation. Hogarth doubts that it captures intuitive processing at a general level.[4]

According to Stanowich & West, an important difference between the two systems is that they tend to lead to different types of task construal. “Construals triggered by System 1 are highly contextual, personalized and socialized. They are driven by considerations of relevance and are aimed at inferring intention and meaning, by the use of conversational implication even in situations that are devoid of conversational features.” The primacy of these mechanisms lead to what has been termed the fundamental computational bias in human cognition. It is a tendency toward automatic and radical contextualization of problems, they argue.[5]

More specifically, this bias includes the tendency to contextualize a problem with as much prior knowledge as is easily accessible, even though the problem is formal and the only solution is a content-free rule. It also includes the tendency to see design and pattern in situations that are random and devoid of pattern and design. Finally, it includes the tendency toward a narrative mode of thought, and toward enthymematical reasoning.[6] This bias then, is called fundamental, and primary, because System 1 is assumed to permeate all of our thinking.[7] In many ways this view resonates well with Jung and the philosophical account where Ideas, Forms, and Archetypes are seen as the laws permeating and “governing the course of all things we can experience.”[8] However, System 1 resembles aspects of the first and second level of intuition, only.

 

 

System 2 on the other hand, de-contextualizes and de-personalizes problems. “This system is more adept at representing in terms of rules and underlying principles. It can deal with problems without social content and it is neither dominated by the goal of attributing intentionality, nor by the search for conversational relevance.”[9]

Stanovich & West argue that System 2 ought to be given priority. This is due to its important function of abstracting complex situations into canonical representations that are stripped of context. This is, the same as the analytic a priori of Kant. “It is likely that one computational task of System 2 is to decouple contextual features automatically supplied by System 1, when they are potentially interfering.”[10] The question we may pose is; how can we separate the subjective brain from its historical and biological context? Also, is it possible to conceive it as separate from its unconscious aspect, be it personal or collective? If such a borderline is illusionary, the distinction between System 1 and 2 may be as well. Much the same concern is voiced by Bargh & Ferguson when they ask; what controls controlled processes?[11] Stanovich & West recognize this problem and write that the override function of System 2 might only be needed in a tiny minority of important information processing situations, and that in most cases the two systems will act in concert. The issue of context is indeed a crucial one. Here it suffices to say that an object or subject devoid of its context necessarily tells us only half the story. It thus eludes a proper rational explanation.

In discussing the two types of task construal Stanovich & West focus in on evolutionary and normative rationality. The biases introduced by System 1 heuristic processing may well be universal – because the computational biases inherent in this system are ubiquitous and shared by all humans, they argue. “However, it does not necessarily follow that, errors on tasks from the heuristics and biases literature will be universal. This is because, for some individuals, System 2 processes operating in parallel will have the computational power to override the response primed by System 1.”[12] Furthermore, they hypothesize that the features of System 1 are designed to very closely track increases in the reproduction probability of genes. System 2, “while also clearly an evolutionary product, is also primarily a control system focused on the interests of the whole person. It is the primary maximizer of an individual’s personal utility. Maximizing the latter will occasionally result in sacrificing genetic fitness.”[13] Thus they argue that because System 2 is more attuned to normative rationality than is System 1, it will seek to fulfill the individual’s goals in the minority of cases where those goals conflict with the responses triggered by System 1. This rather cryptic trade-off between personal utility and genetic fitness is not elaborated by Stanovich & West and several authors oppose them on this issue.[14]

Being aware of this intricate issue, they stress that “in the vast majority of mundane situations, the evolutionary rationality embodied in System 1 processes will also serve the goals of normative rationality. Our automatic, System 1 processes for accurately navigating around objects in the natural world were adaptive in an evolutionary sense, and they likewise serve our personal goals as we carry out our lives in the modern world.”[15] Interestingly, Stanovich & West argue that one way to view the difference between evolutionary and normative rationality is that they are not really different types of rationality. Rather, “they are terms for characterizing optimization procedures operating at the sub-personal and personal levels, respectively.”[16]

In applying these notions, we are reminded that Jung anchors his criterion of analytical judgment in the personal ego. I suggested that a valid criterion for intuitive judgment is to be found in the self, which includes both the personal and the sub-personal domain. This is also in accordance with the philosophical account. Thus, we may have come full circle. However, the appealing promise of the third level of intuition is of a unifying and integral consciousness. It is the result of a thorough and simultaneous induction and deduction, perceiving the a priori and the a posteriori, the personal and sub-personal as one. In summarizing then, I agree with March and suggest that both systems serve the fulfillment of an identity, and both are therefore intrinsically subjective.[17] Perhaps neither one provides the rationale for a normative theory of rationality. And perhaps normative rationality is ontologically dependent on intuition, which will be advocated in the next chapter.

 

 

Efficacy of Analytical vs. Intuitive Thinking in Expert Judgment

As this thesis focuses on analytical and intuitive thinking in strategic decisions, it is of interest to look at efficacy. We will return to this issue in the succeeding chapter on intuition in strategy. Here I limit myself to a discussion of a particularly interesting study undertaken by Hammond, Hamm, Grassia and Pearson.[18] Corresponding to the increased concern with unconscious or implicit phenomenon within cognitive psychology, experimental interest in intuitive sources of judgment has increased. Hammond et al. write that intuition is frequently assumed to be the basis for judgments made rapidly and easily, without awareness of the inferences supporting them. Eisenhardt & Zbaracki state that: “studying intuition is a way to create a more realistic view of how strategic decision makers actually think.”[19]

Performance based on intuitive judgments of the correct solution to a problem, are typically compared with some established scientific procedure for arriving at a solution. Much of the research of this nature, particularly in earlier studies, focuses on the shortcomings of intuition in comparison with analytical processing.[20] This is also the case with the well-known study: Clinical versus Actuarial Judgment, by Dawes, Faust, and Meehl. In considering factors underlying the greater accuracy of actuarial methods, they emphasize that the mathematical features of actuarial methods “ensure that variables contribute to conclusions based on their actual predictive power and relation to the criterion in interest.”[21] They also warn that clinical judgments produce self-fulfilling prophecies. Or as Kahneman and Tversky write: “The prevalent tendency to underweight or ignore distributional information is perhaps the major error of intuitive prediction.”[22]

In this type of research, the rationality of a person’s intuitive judgment under uncertainty is usually compared with analytically derived answers produced by a formal model such as Bayes’s theorem, a multiple regression equation, or other rules from the conventional probability calculus.[23] A key point addressed by Hammond et al., is that such comparisons are indirect: they compare a person’s intuitive efforts with person-independent operations. That is, they compare a person’s intuitive processes and judgment with those of an analytically derived rule or equation put forward as a standard of rationality. Such a comparison cannot provide a test of whether analytical cognition is inferior or superior to intuitive cognition, and under what conditions.

Indirect comparisons are undeniably important, but they are necessarily restricted in three ways, as Hammond et al. argue. “First, because indirect comparisons evaluate intuition with respect to a standard of rationality, researchers must choose one standard from the many offered. However, agreement on which standard of rationality is correct, has never been achieved. The choice of any standard, therefore, is subject to dispute, and any conclusion that subjects have failed to achieve the standard chosen are sure to be criticized by those who prefer a different standard.”[24] The difficulties in comparing intuitive and analytical judgments are thus left unresolved. This argument is indeed a pivotal one, and it serves to legitimate the relevance of both the theoretical and empirical part of my work. It is further elaborated in the next chapter on intuition and rationality.

Along the same line of reason a second argument is developed. They point to the fact that indirect comparisons, “cannot fail to show that analytical cognition is equal or superior to intuitive cognition because analytical models, however chosen, provide a ceiling for performance.”[25] It is thus not surprising, they argue, that studies find that few persons’ intuitive efforts achieve the standard, and none exceed it. Finally, “when indirect comparisons are made, the analytical models are always provided with all the correct (and only the correct) substantive information each model requires.”[26] They note that in journals, such models are usually executed without error. “In practice, however, the analytical cognition of persons, in contrast to analytical computation by formal models, is vulnerable to substantive failures (insufficient information, incorrect information, incorrect substantive theory) and to procedural failures (incorrect assignment of numbers to the symbols of the equation, computational errors, use of an incorrect model, insufficient time).”[27] Yet another obstacle is the conceptual problems, elaborated in this thesis. Despite a long history of dispute then, these concerns are usually not addressed when the efficacy of intuitive and analytical cognition is compared. Because the research of Hammond et al. is the first scientific effort to compare them directly, their results are pioneering.

Indirect comparisons may indeed be valuable, but the restrictions described by Hammond et al., prevent them from informing us about the relative efficacy of the intuitive and analytical cognition of strategists.[28] Direct comparisons on the other hand, will reveal the efficacy of these modes of cognition in terms of empirical achievement or correctness. Comparisons of relative efficacy, however, require the presence of an empirical criterion with which judgments are compared, rather than a standard of rationality. The comparison undertaken in the empirical part of my thesis is direct, and anchors the relative efficacy, in empirical decision quality criteria. It is the strategist that can and should define these criteria, which may differ from case to case. When a set of criteria is available, direct comparisons enable us to address the age-old question: does a person’s intuitive or analytical cognition produce the more empirically favorable and accurate answer?

Before we look at their main findings, we should take note of the definitions applied. They argue that cognitive processes can be arranged on a continuum that runs from intuition to analysis, and that any point on such a continuum interact in a predictable way with various task conditions located on a similar continuum.[29] Their list of task conditions is included below, as it is of relevance to the explorative and exploitative decision making contexts applied in my empirical study. They also say that the task properties induce intuitive or analytical cognition in order to avoid implying that the relation between task properties and cognitive properties is inevitable or fully deterministic.

 

 

Task Conditions Inducing Intuitive and Analytical Cognition:

Cognitive   Processes

Intuition

Analysis

Cognitive control Low High
Rate of data processing Rapid Slow
Conscious awareness Low High
Organizing principle Weighted average Task specific
Errors Normally distributed Few, but large
Confidence High in answer, low in method Low in answer, high in method

 

 

 

Task Conditions

Intuition-inducing state of task characteristics Analysis-inducing state of task characteristics
Number of cues Large > 5 Small
Measurement of cues Perceptual measurement Objective reliable measure
Distribution of cue values Continuous Discrete
Redundancy among cues High redundancy Low redundancy
Decomposition of task Low High
Degree of certainty in task Low certainty High certainty
Relation between cue and criterion Linear Non-linear
Weighting of cues in environmental model Equal Unequal
Availability of organizing principle Unavailable Available
Display of cues Simultaneous display Sequential display
Time period Brief Long

 

 

 

When direct comparisons were made of the efficacy of 21 expert highway engineers’ use of intuition and analysis, it was found that not only can intuitive cognition perform as well as analytical cognition, but it can outperform it as well. Secondly, they found that analytical cognition is more likely than intuitive cognition to produce extreme errors. This latter result is in agreement with research done by Peters et al. who found that the analytic approach to problem solving produces precise answers more often, but the distribution of errors is quite wide. In contrast to this, intuition is less frequently precise, but more consistently close to the correct answer.[30] Hammond et al. found that the greater the correspondence between task properties and cognitive properties, the greater the subject’s achievement. “This result indicates the task circumstances for which each form of cognition is likely to be most efficacious and therefore appropriate.”[31]

Here we should note that a strategic decision making situation is typically characterized by more intuition-inducing task characteristics. It is a point of some importance for the analysis of my empirical findings. The latter conclusion of Hammond et al. is the more controversial because it contradicts the predominating argument that intuition produces biased incorrect judgments, and thus should be replaced by analytical methods. “In our view, the contradiction is not an artifact. It occurs for two reasons: (a) the tendency to use multiple definitions of intuition and (b) the reliance on indirect comparisons, between persons and equations, for example, rather than on direct comparisons of intuition and analysis within persons.”[32] A suggestion for further research is hence to focus in on conceptual development and direct comparison, which is the ambition in this thesis.

In concluding this brief discussion of dual-process theories I agree with Bargh & Ferguson who argue that conscious and non-conscious processes presumably act in concert, and that both System 1 and 2 are automatic, and determined processes. Furthermore, they emphasize that previously, automaticity has been taken as evidence that the cognition is environmentally determined. This is questioned and they recommend that future research should aim at discovery and delineation of the mechanisms that control controlled processes.[33] Throughout this thesis, I have indicated that the direction to look is towards the personal and collective unconscious. That is, towards the rational and intelligible world of Ideas, Forms, and Archetypes, which represent the a priori laws “governing the course of all things we can experience”, to quote Jung.[34] Moreover, the claim is that it is intuition that provides insight into how this domain is integral to individual consciousness, causing it to work in a more or less dual and determined manner.

 

 

 

 

Summary

 

Philosophers define rational intuition as superior to discursive thinking. In modern psychology, the situation is more or less reversed. Here the tendency is towards one, which equals intuition with preconscious incubation, emotional involvement, subjective consistency, automatic, rapid, biased and effortless processing. It is rather obvious from this psychological account that the reference to the philosophers’ intelligible world of pure reason has more or less disappeared in the great black void of the personal unconscious and in the issue of normative rationality. There is hardly any discussion of the universal Ideas, Forms, Archetypes and laws that may govern it. In this chapter, I have painstakingly and at great length aimed at delineation of how this Copernican reversal in our history of epistemology has taken place. The fragmentation that is sneaking in causes us to wonder how intuition relates to rationality. This issue is not easily settled but will be further elaborated in the next chapter. There it is argued that intuition is the ontological foundation for any theory of normative rationality.

 

 


[1] The Neoplatonists defined it as inferential or non-inferential.

[2] Stanowich & West, 2000, p. 658-659. See also Gilovich et al., 2002, p. 51, 379, 436.

[3] Ibid. My italics. They refer to Levinson, 1995. See also Kahneman et al., 1985.

[4] Epstein et al., 1996, p. 392. Hogarth, 2001, p. 268.

[5] Ibid. p. 659. See also Hogarth, 2001, p. 268.

[6] Ibid. Enthymematical reasoning is defined as making assumptions not stated in a problem and then to reason from those assumptions.

[7] Ibid. p. 662.

[8] Jung, 1971, p. 401.

[9] Stanowich & West, 2000, p. 659. See also March, 1994, p. 57.

[10] Ibid. p. 662. They refer to Navon, 1989.

[11] Bargh & Ferguson, 2000, p. 938. In yet other words, we may question whether or not analyzing a distorted dual mind will take us any closer to ‘truth’. See also Sternberg, 1994.

[12] Stanowich & West, 2000, p. 660-661.

[13] Ibid.

[14] Oberauer, 2000, p. 692, Newstead, 2000, p. 690, in Stanowich & West, 2000. S&W refer to Dawkin 1976, who see evolutionary adaptation as the optimization process of the genes, whereas normative rationality concerns utility maximization for the so-called vehicle, which houses the genes.  

[15] Ibid. p. 661.

[16] Ibid.

[17] March, 1994, p. 61.

[18] Hammond, et al. 1997, p. 144-174.

[19] Eisenhardt & Zbaracki, 1992, p. 33. See also Lieberman, 2000, p. 109, Baylor, 2001, p. 237-243, Bargh & Ferguson, 2000, March, 1994, p. 262, Markley, 1988, p. 85, Schooler, et al., 1999, p. 280, Simon, 1997, p. 129, Eisenberg, 1984, p. 85, Mintzberg, 1994, p. 303, Hill, 1988, p. 137, Buckingham, 2000, p. 990, Hamm, 1988, p. 78, Eisenhardt, 1999, p. 65, Andersen, 2000.

[20]See the work of Slovic, Fischhoff, Lichtenstein, Kahneman, Tversky.

[21] Dawes, Faust, and Meehl, 1985, p. 1671.

[22] Kahneman, et al. 1985, p. 416.

[23] Hammond, et al. 1997, p. 144. They refer to Einhorn, 1981, Jungerman, 1983, Kahneman, Sovic & Tversky, 1985, and Pitz, 1984.

[24] Ibid. p. 144, 171.

[25] Ibid.

[26] Ibid. They refer to Kahneman & Tversky, 1982, and Wright & Murphy, 1984.

[27] Ibid.

[28] Ibid.

[29] Ibid. p. 146-149.

[30] Peters, 1974, p. 125-131. See also Weigelt, 1988.

[31] Hammond et al. 1997, p. 172.

[32] Ibid.

[33] Bargh & Ferguson, 2000, p. 939-941.

[34] Jung, 1971, p. 400-401. My italics.