Abstract Beyond General Intelligence: The DualProcess Theory of Human Intelligence Scott Barry Kaufman 2009 Over 30 years of research in cognitive science reveals that a considerable amount of information processing takes place automatically—without our intent, awareness, or deliberate encoding—and plays a significant role in structuring our skills, perceptions, and behavior. Indeed, it is increasingly recognized that dual‐process theories, which posit that humans possess two distinct modes of thought—one controlled, and the other automatic— are required for explaining cognitive, personality, and social phenomenon. However, while intelligence researchers have done a remarkable job measuring individual differences in explicitly controlled cognitive processes, individual differences in automatic cognitive processes have not received nearly as much attention. In this dissertation, I aim to go beyond general intelligence (g) by measuring individual differences in implicit cognition and their relations to a wide variety of intelligent behaviors, thereby expanding both the range of methodologies as well as dependent measures studied by intelligence researchers.
Toward these goals, I proposed the Dual‐Process (DP) theory of human intelligence in which intelligent behavior is jointly influenced both by Controlled and Autonomous forms of cognition. According to the DP theory, intelligence is the ability to balance and flexibly switch between modes of thought depending on task demands. While Controlled Cognition is largely constrained by central executive functioning, Autonomous Cognition is not. Further, both ability and Openness to Engagement in Autonomous forms of Cognition are expected to predict a wide variety of intelligent outcomes independently of Explicit Cognitive Ability (ECA).
The theory was largely supported in a sample of 177 English Sixth Form College students between the ages of 16‐18. Two forms of implicit cognition tested were implicit learning (IL) of a probabilistic sequential pattern and latent inhibition (LI) of stimuli that was previously tagged as irrelevant. IL and LI were both unrelated to measures of ECA (g, working memory, and intentional associative learning) and Intellectual Engagement. Yet IL and LI both displayed meaningful individual differences.
IL was positively related to specific components of cognitive ability, Openness to Experience, impulsivity, and language achievement. Reduced LI was positively associated with Openness to Affective Engagement and self‐reported creative achievement in the Arts, but not the Sciences. Additionally, Openness to Affective and Aesthetic (Art, Music and Fantasy) Engagement differentially predicted deductive reasoning and self‐report measures of the Big Five (Costa & McCrae, 1992), impulsivity, need for uniqueness, and creative achievement in the Arts above and beyond ECA and Intellectual Engagement. ECA and Intellectual Engagement were related to self‐reported creative achievement in the Sciences, but not the Arts.
Taken together, these results have implications for understanding rationality, evolutionary psychology, interactions between Controlled and Autonomous Cognition, social cognition, creativity, schizophrenia, expertise, and the relationship between personality and cognition. As such, these results illustrate how—by investigating individual differences both in Controlled and Autonomous forms of cognition—the Dual Process theory provides a more complete understanding of human intelligence. Beyond General Intelligence: The DualProcess Theory of Human Intelligence A Dissertation Presented to the Faculty of the raduate School of Yale niversity in Candidacy for the Degree of Doctor of Philosophy by Scott Barry aufman Dissertation Directors eremy R. Sternberg May, 2009 P a g e | ii Copyright 2009 by Scott Barry aufman All rights reserved.
P a g e | iii Table of Contents Chapter Page # Acknowledgements. Methodology 68 Part I Ability 3. Explicit Cognitive Ability 89. 1 9 Part II – Engagement 6.
Four‐Factor Model. 261 … References……………………………………………………………………………………………………… 28 Appendix A Questionnaires……………………………………………. 3 8 Appendix BAdditional Covariance Analyses……………………………………………………. 37 Appendix C Additional Implicit Learning Tasks……………………………………………….
380 P a g e | iv List of Figures Page # Figure 11. The DualProcess (DP) Theory of Human Intelligence 26 Figure 21. Representation of the procedure used for the probabilistic SRT 77 learning task Figure 31. Associative learning, working memory capacity (WMC), and 102 processing speed (Gs) independently predict g (N=169) Figure 41.
SRT Learning Performance for probable (SOC85) and non 125 probable (SOC15) trials across one practice and eight learning blocks (N = 153) Figure 42. Associative learning (AL), working memory (WM), processing speed 132 (GS), and explicit cognitive ability (g), and verbal reasoning (DATV) predict implicit learning (IL) (N= 153) Figure 43. Double dissociation between Openness and Intellect in predicting 137 working memory (WM) and implicit learning (IL) (N = 153) Figure 51. Bimodal distribution of latent inhibition scores in the preexposed 153 condition (N 121) Figure 52.
Interaction between faith in intuition (FII) and mean number of 160 trials to correct rule identification in the preexposed and nonpreexposed conditions Page |v Figure 81. Distribution of Creative Achievement Questionnaire (CAQ) scores 209 (N=177) Figure 82. Selfreported Arts and Sciences Achievement scores of highlow 222 latent inhibition (LI) and highlow Explicit Cognitive Ability groups (N= 97) Figure 91. Social Exchange Scenario 241 Figure 92.
(a) Mean proportion correct and (b) mean response time by 243 condition with bars representing S. of the mean. The finitestate grammar used in the current dissertation. The 383 grammar generates letter strings by following the arrows from the input state (s1) to the terminal state (s6).
The difference between cued (C) and random (R) trials 389 Figure C3. Contextual Cueing Performance for fixed and variable trials 391 (N = 175) P a g e | vi List of Tables Page # Table 11. Dualprocess theories 10 Table 12. Properties of the Two Systems 15 Table 21.
Order of Test Administration 70 Table 31. Descriptive statistics for learning trials on 3Term and PA (N = 169) 97 Table 32. Correlations, means, and standard deviations of observed variables 99 (N = 169) Table 33. Correlation matrix of latent variables in structural model (N = 169) 101 Table 41.
Correlations among all measures of g, ECTs, IL, Intellect, Openness, 128 Intuition, and Impulsivity Table 42. Correlations among implicit learning and latent variables for g and 130 ECTs (N 1 3) Table 43. Correlations between GCSE scores and g, ECTs, and implicit learning 134 Table 44. Correlations between cognitive tasks, latent variables for Intellect 136 and Openness, and Intuition Table 51.
Correlations among REI and MBTI subscales and LI scores 152 Table 52. Factor analysis of REI experiential items 155 Table 53. Correlations among REI experiential factors and MBTI subscales. Factor Analysis of REI experiential factors and MBTI subscales (N = 158 163) Table 55.
Correlations of g and associated ECTs with Latent Inhibition 161 P a g e | vii Table 61. Factor Analysis of all Explicit Cognitive Ability, Intellect, Openness to 175 Experience and Intuition measures (N = 146) Table 62. Top 10 Loadings on a principal component consisting of all REI 176 Rational Favorability, NEO Ideas, and BFAS Intellect items Table 63. Top 10 Loadings on a principal component consisting of all NEO 177 Feeling, MBTI Feeling, and REI Experiential items Table 64.
Top 10 Loadings on a principal component consisting of all NEO 178 Aesthetics, NEO Fantasy, BFAS Openness, and MBTI Intuition items Table 65. Correlations among the four factors (N = 146) 180 Table 71. Correlations between the fourfactor model of cognitive traits and 183 Neuroticism, Agreeableness, Conscientiousness, and Extraversion (N= 143) Table 72. Correlations among the fourfactor model of cognitive traits and the 184 aspects of Neuroticism, Agreeableness, Conscientiousness, and Extraversion (N= 143) Table 73.
Correlations between the fourfactor model of cognitive traits and 189 fourfactor model of impulsivity (N= 145) Table 74. Factor analysis of Need for Uniqueness items (N=113) 195 Table 75. Correlations between fourfactor model of cognitive traits (N=112) 197 and need for uniqueness Table 81. Spearman’s rho correlation among fourfactor model of cognitive 208 traits and selfperceived talent in 13 domains (N= 146) P a g e | viii Table 82.
Spearman’s rho correlation among the fourfactor model of 211 cognitive traits, 10 domains of achievement, and total creative achievement (N= 146) Table 83. Factor Analysis of REI experiential factors and MBTI subscales (N = 214 177) Table 84. Spearman’s rho correlations among the fourfactor model of 215 cognitive traits, and selfreported achievement in the Arts and Sciences (N=146) Table 85. Spearman’s rho correlations among need for uniqueness, and self 226 reported achievement in the Arts and Sciences (N=113) Table 86.
Correlations between the fourfactor model of cognitive traits and 227 fourfactor model of impulsivity (N= 146) Table 91. Correlations between Accuracy of Deductive Reasoning and g, ECT’s, 245 Intellectual Engagement, Affective Engagement, and Aesthetic Engagement Table 92. Correlations between speed of deductive reasoning and g, ECT’s, 247 Intellectual Engagement, Affective Engagement, and Aesthetic Engagement Table B1. Full covariance matrix used to fit SEM model in Chapter 3: Explicit 376 Cognitive Ability (N = 169) Table B2.
Full covariance matrix used to extract fourfactor model in Chapter 378 6: FourFactor Model Table B3. Factor Analysis of BFAS and UPPS scales (N = 160) 380 Table B4. Factor Analysis of BFAS and UPPS Scales (N=160) 381 P a g e | ix Table C1. Correlations among the 4 implicit learning tasks administered in the 394 current dissertation Page |x Acknowledgements While dramatic, it wouldn t be an understatement to say that this dissertation is the culmination of my life s work up to this point.
Ever since I can remember, I ve been deeply fascinated with variations in human intelligence. I am indebted to many people along the way who have contributed, in one way or another, to the opportunity for me to complete this dissertation. First, my friends. Two chaps who have been particularly important in my life are Elliot Samuel Paul (my twin brother from another mother) at Yale niversity and Dr.
Benjamin Irvine who I first met at niversity of Cambridge while completing my Masters degree. Out of coincidence, or perhaps some cosmic reason I am not (yet) consciously aware, they are both philosophers. Thanks to both of them for their friendship, the many stimulating conversations and for helping me to maintain some semblance of a social balance in my life. Many thanks must also go to other friends of mine whom I ve been honored to know throughout the years Eugene Ford, Markus LaBooty, ienke enderbosch, Brent yle, ennifer DiMase, Louisa Egan, ane Erickson, Elise Christopher, Candida Moss, Ruhl Due as, Erin Coulter, ustin hoo, Mark erban, Avi ouzi, amie Brown, and Balazs Aczel.
I also owe a great deal of my inspiration to the members of the Bret Logan discussion group Bret Logan, Alia Crum, Yoona ang, Dave Roberts, and Adam reen. ext, my collaborators. A huge debt of gratitude must go to the following individuals Colin DeYoung (whose statistical help, general guidance, and friendship truly made this dissertation possible), ames C. As for my lifelong mentors, warm appreciation goes to Anne Fay and icholas Mackintosh for their continual guidance.
Thanks to the late Herbert Simon and Randy Pausch for mentoring and inspiring me as an undergraduate and to eremy ray for giving me a home at Yale. Warm appreciation goes to my High School teachers Mary Acton, Regina ordon, Tom Elliot, and Debra Hobbs for their fine teaching and unbridled encouragement and Mr. O for his stimulating and fun creative writing class. I also owe gratitude to various individuals who provided me with the additional support necessary to bring this dissertation to fruition.
First and foremost, a huge amount of good cheer and thanks must go to Hills Road Sixth Form College in Cambridge, England for their repeated willingness to allow me the use of their facilities as well as allowing me to run psychology experiments on their students. Sheila Bennett has been immensely helpful in assisting with the recruitment of participants. Both im Blair at Hills Road and ikhil Srivastava at Yale niversity have been extremely helpful with computer support. Thanks to Leib P a g e | xii Litman, Ben Williams, Stephen Pearlberg, and Shelley Carson for providing me with various testing materials.
Deidre Reis was kind enough to allow me to adopt her materials for the computerized deduction reasoning task. My appreciation also goes to Arthur Reber for lively discussion and valuable input.