p.15
2
VIRTUAL PERSONALITIES
A Neural Network Model of the Structure and Dynamics of Personality
Stephen J. Read, Vita Droutman, and Lynn C. Miller
There is a long-standing split in personality between structural and dynamic approaches to understanding personality (for recent discussions, see Funder, 2001; Mischel & Shoda, 1998). These two approaches typically rely on different theories and constructs. The structural approach typically focuses on determining the psychometric structure of personality tests (e.g., Lee & Ashton, 2004; McCrae & Costa, 1999) or the everyday language used to describe personality (Digman, 1997; Goldberg, 1981; John & Srivastava, 1999), and a major (although not the only) focus is on understanding stable personality dispositions. In contrast, dynamic approaches, such as Mischel’s Cognitive-Affective Systems theory (Mischel & Shoda, 1995, 2008), Atkinson’s Dynamics of Action model (Atkinson & Birch, 1970) or McAdams’ narrative approach to personality (McAdams, 2008; McAdams, Josselson, & Lieblich, 2006) typically focus on uncovering and understanding the psychological mechanisms that underlie individual differences and changes in behavior across situations and time, and are less interested in understanding stable, broad dispositions. Although some personality researchers have suggested that there is no need to try to bring these two approaches together, many researchers have become increasingly concerned with doing so.
In the current chapter we argue that modeling personality in terms of the behavior of organized motivational systems can help unify the two approaches. We present a neural network model of personality in terms of structured motivational systems (Read et al., 2010; Read & Miller, 2002) that both demonstrates how the structure of personality, such as the Big Five, can arise from these motivational systems and how the dynamics of everyday behavior can arise from the interaction between these structured motivational systems and the changing affordances of situations over time and the bodily states of individuals. Our goal is to show how personality structure and personality dynamics can arise from the same psychological architecture in interactions with the social and physical environment.
p.16
People differ considerably in their behavior and how they respond to the same situations. Personality psychologists who take the psychometric approach to personality often try to understand these individual differences by asking people to respond to items on a personality test, which typically ask people to characterize themselves in terms of typical behaviors, beliefs, emotions, and sometimes motivation. The response to these items across a large sample of individuals is factor analyzed, and the result is typically something like the Big Five (John, Naumann, & Soto, 2008; John & Srivastava, 1999) or similar structures such as the six-factor HEXACO model (Ashton & Lee, 2007; Lee & Ashton, 2004), which essentially adds an Honesty/Humility factor to the Big Five. The Big Five (often referred to by the acronym OCEAN) consists of five major dimensions of human personality: Openness to experience, Conscientiousness, Extraversion, Agreeableness, and Neuroticism.
Differences between individuals can then be considered in terms of different individuals’ relative standing on each of the Big Five dimensions. Each individual can be thought of as having a relative profile across each dimension that describes their standing relative to others on the various dimensions.
But something is typically missing in the various structural approaches to personality: an explanation of why this particular structure is found and what are the underlying mechanisms or processes. We have argued that this particular structure is the result of underlying structured motivational systems.
Our neural network model of personality (Read et al., 2010; see also Read & Miller, 1989) argues that personality traits, and specifically the Big Five, arise from the interaction between structured motivational systems and the goal affordances of situations. We have argued that a number of relatively specific brain systems manage a variety of different motivational domains and their related behavior, and that these specific systems are then organized into two higher-level Approach and Avoidance systems that integrate over the lower-level systems. The structure of these motivational systems is responsible for the structure of human personality. But additionally, the dynamics of personality can be understood in terms of the ways in which these structured motivational systems interact with the affordances of different situations that the individual encounters and with the individual’s current bodily state.
As noted above, personality researchers have largely addressed the structure and dynamics of personality independently. However, researchers have become increasingly interested in how those two important approaches can be related. Here we use our neural network model of personality to show how the structure and dynamics of personality can arise from the dynamics of structured motivational systems.
In the first section of the chapter, we demonstrate how such structured motivational systems can transform variability in chronic motivations across individuals into patterns of behavior from which we can recover the Big Five. We will do this by constructing a neural network model of a set of structured motivational systems, and then training the network to associate different clusters of behavior with different underlying motives. This network will then be used to simulate the behavior of a large number of different individuals as they respond to a variety of different situations. We will then factor analyze the resulting behavior. The result of this factor analysis will map onto the underlying motivational structures in the model.
p.17
In the second section of the chapter, we will show how both within- and between-person variability in personality-related behavior can be understood in terms of the dynamics of the interaction between individual’s motives, the affordances of situations, and current bodily states. In a series of recent experience-sampling studies of everyday behavior, Fleeson has shown that while there are clear individual differences, consistent with the Big Five, in trends in personality related behavior over time, at the same time the within person variability over situations is at least as large as the between person differences (Fleeson, 2012; Fleeson & Jayawickreme, 2015).
In the final section of the chapter, we will discuss the implications of our computational model for thinking about person–situation interactions.
A number of researchers (e.g., Bugental, 2000; Kenrick & Trost, 1997) have noted that there are a wide range of problems and tasks that all humans must solve if they are to survive and successfully reproduce, and they have identified a number of motivational systems that have evolved to handle these tasks (e.g., Maslow, 1943; Murray, 1938). Among the various tasks that must be solved and the relevant system are mating, nurturance of young, affiliation and bonding with peers, establishing dominance hierarchies, insuring attachment to caregivers, avoidance of social rejection, and avoidance of physical harm. Over the last 15 years we have developed two extensive motive taxonomies from which we draw in our work (Chulef, Read, & Walsh, 2001; Talevich, Read, Walsh, Chopra, & Iyer, 2015).
In addition, work from neuroscience and work on biological temperament has demonstrated that there are at least three major biologically based dimensions of personality or temperament that map onto three dimensions of the Big Five: Extraversion, Neuroticism, and Conscientiousness. Each of these three dimensions corresponds to a major aspect of motivation. A number of researchers have argued that there is a general Approach (Clark & Watson, 1999; Rothbart & Bates, 1998) or Behavioral Approach System (BAS) (Gray, 1987; Gray & McNaughton, 2000) that governs sensitivity to cues signaling rewards and, when activated, results in an active approach. This maps onto Extraversion. There is considerable evidence that the neurotransmitter dopamine plays a major role in this Approach or Reward system.
In addition, there is a general Avoidance or Behavioral Inhibition System (BIS) (Carver & White, 1994; Clark & Watson, 1999; Gray, 1987; Gray & McNaughton, 2000) that governs sensitivity to cues of punishment and loss and manages avoidance of threatening situations. This maps onto Neuroticism. A third broad dimension concerns a General Disinhibition/Constraint System (Derryberry & Rothbart, 1997; Rothbart & Bates, 1998) which is related to executive function and inhibitory control. This probably maps onto Conscientiousness.