A second key claim of the massive reuse model is that the same affective systems that operate during our ordinary transactions with the world and that deliver ongoing evaluative information are reused during deliberation. How could decision-making work if it were not true? In Chapter 2, we saw that affective systems maintain a rich and detailed map of the evaluative landscape. These systems use sophisticated statistical machinery to maintain and rapidly update representations of the goodness or badness of situations, allowing them to intelligently shape behavior in real time. For example, if a person is on a quiet beach with a bright sun overhead and cool drink at hand, this person's affective system rapidly registers this as a very good situation. There is an abundance of positive affect, what might be described as an affective "inner glow."
It is entirely natural to think that during deliberation when people bring to mind the prospect of going to the beach versus an alternative (say, staying at home), these very same affective reactions are called up as a way of registering how good or bad these prospects are. When they think about going to the beach, they experience a bit of this same inner glow that they would have experienced had they actually been at the beach, and this helps to explain why they chose the beach. This picture fits with the phenomenology of deliberation and the common sense understanding of how deliberation works.
The alternative model, the separate processors view, is that there is some other entirely distinct set of "cool" representations of value that come on line during deliberation. Affective systems guide ongoing action in real time but with a different set of nonaffective representations coming on line during slow, serial, deliberative reflection. This picture is rarely explicitly articulated by separate processor theorists, but nonetheless, it seems implicit in the model. For example, a model along these lines appears to be suggested by Kahneman in his discussion of the affect heuristic, which he calls "probably the most important development in the study of judgment heuristics in the past few decades" (Kahneman, 2003). Drawing on the work of his colleague Paul Slovic and others, Kahneman says that when presented a hard choice, people often use their quick affective reactions as a guide to what to do. This is contrasted with the alternative process of going through a slower and more difficult "analytic" cost-benefit calculation (Slovic, Peters, Finucane, & MacGregor, 2005). Neither Kahneman nor Slovic say how this cost-benefit calculation works in any detail, but if it is supposed to be a genuinely different way to arrive at a judgment than consulting one's affect, then one must assume that it is a process that is nonaffective. That is, it must use some other kind of "cooler" representation of value.
Kahneman's picture—if we understand it correctly—is implausible. Why would the mind be built this way with both rich and detailed hot representations of value as well as a separate set of cool, nonaffective representations of value? There is already an elaborate and sophisticated system designed for tracking the value of actually encountered situations: the affective system. Wouldn't it be wasteful (and redundant) to cast this system aside during deliberation and instead rely on some other system to track the value of prospective situations?
There is neurobiological evidence that supports the idea of affective redeployment that is central to the massive reuse model. Antonio Damasio, Antoine Bechara, and their colleagues have pursued a longstanding research program examining patients with selective damage to discrete regions of the brain as a way to illuminate the role of affect in decision-making.
In one line of research, they investigated the Iowa Gambling Task. This is a task in which participants make repeated selections from four decks of cards (Bechara, Damasio, Tranel, & Damasio, 1997, 2005; Damasio, 1994). With each card selected, participants receive a potential reward (gain of money) or punishment (loss of money). Unbeknownst to the participants, two decks have relatively favorable payoffs ("good" decks) while two are unfavorable in that they have initially high rewards but thereafter have even larger punishments ("bad" decks). Healthy participants rapidly learn to move away from the bad decks and select from the good ones and they appear to learn to do this mostly nonconsciously. Skin sensors that measure emotional responses reveal that participants rapidly acquire quick, spontaneous affective responses, what are often called "gut feelings," as they make choices from the decks, and these affective cues guide them away from the bad decks and toward the good ones. These affective reactions emerge quite early in the task, well before participants can explicitly articulate why they are favoring certain decks and avoiding others.
Damasio, Bechara, and their colleagues found that patients with damage to their ventromedial prefrontal cortex (vmPFC), a region that houses important affective valuation circuits, do not generate anticipatory affective reactions as they perform the task. Moreover, they fail to switch away from the bad decks. They lack knowledge of the costs or benefits of the various decks and continue to select from all decks—good and bad—alike. Of course, these results should be not at all surprising to our readers because they fit nicely with our argument from Chapter 2 that affective systems are sensitive to hidden patterns of rewards and punishment. These systems use sophisticated statistical routines to rapidly calculate and update the goodness or badness of the person's options, thus providing critical ongoing intuitive guidance for action.
If the separate processors view were correct, then the problems of these patients would be relatively specific to contexts requiring affect to provide rapid intuitive guidance of action. When these patients step back and engage in slow, effortful deliberation about what to do, then a separate processor with a distinct set of "cool" nonaffective value representations should come on line. In contrast, the massive reuse view says that when affective systems are compromised, then not only is intuitive guidance by affect impaired, but deliberative guidance will also be impaired because it makes massive reuse of affect.
Damasio and colleagues' observations strongly support the predictions of the massive reuse view. The famous case of Phineas Gage provides one illustration. After a tamping iron pierced his skull, destroying his vmPFC, he subsequently became impulsive, inappropriate, and made a series of disastrous decisions (Damasio, 1994). Gage, of course, was a historical figure and surely has been subject to the accumulation of distortions and hyperbole. Damasio and his colleagues were able to undertake comprehensive neuropsychological investigation of other patients with vmPFC damage (Damasio, 1994). This includes studies of Elliot, a patient with selective damage to the vmPFC established by modern day radiographic imaging.
Comprehensive testing showed that Elliot's general intellectual abilities were unimpaired:
The standardized psychological and neuropsychological tests revealed a superior intellect. On every subtest of the Weschler Adult Intelligence Scale, Elliot showed abilities that were either superior or average. His immediate memory for digits was superior, as were his short-term verbal memory and visual memory for geometric designs.
His performance on the Multilingual Aphasia Examination, a battery of tests which assess various aspects of language comprehension and production, was normal. His visual perception and construction skills were normal on Benton's standardized tests of facial discrimination, judgment of line orientation, tests of geographic orientation, and two- and three-dimensional block construction.... In short, perceptual ability, past memory, short-term memory, new learning, language, and the ability to do arithmetic were intact (Damasio, 1994, p. 41).
Despite this, Elliot's decision-making was seriously compromised.
His knowledge seemed to survive, and he could perform many separate actions as well as before. But he could not be counted on to perform the appropriate action when it was expected. Understandably, after repeated advice and admonitions from colleagues went unheeded, Elliot's job was terminated. Other jobs—and other dismissals—were to follow . No longer tied to regular employment, Elliot charged ahead with new pastimes and new business ventures.. In one enterprise, he teamed with a disreputable character. Several warnings from friends were of no avail, and the scheme ended in bankruptcy. All of his savings had been invested in the ill-fated enterprise and were lost. It was puzzling to see a man with Elliot's background make such flawed business and financial decisions. His wife, children, and friends could not understand why a knowledgeable person who was properly forewarned could act so foolishly ... (Damasio, 1994, pp. 36-37).
According to Damasio's descriptions, Elliot's social world also began to disintegrate. He left his wife, entered into several unwise relationships, had several subsequent divorces, and ended up drifting from place to place without an income.
Based on the experiments with the Iowa Gambling Task, neurological case studies such as Elliot, as well other lines of evidence, Damasio proposed the somatic marker hypothesis (Bechara & Damasio 2005; Damasio, 1994). This is a complex model, but for our purposes, the most relevant point is that Damasio draws a deep connection between affect and deliberation. He proposes that affective systems are called on during consideration of options, biasing action away from bad prospects and toward good ones. It is precisely this affective guidance during contemplation of what do that is absent in Gage and in Elliot, accounting for their disastrous decision-making.
The idea that affect is redeployed during deliberation also finds support from neuroimaging studies. Most imaging studies have examined rapid, on line selection of action; that is, in our terminology, they probed intuitive guidance. In these studies, participants are presented with dozens, sometimes hundreds, of choices and have a very brief period, usually just 2 to 3 seconds, to respond to each choice. Areas known to be implicated in intuitive, affective processing— vmPFC regions—are reliably activated, and they appear to supply evaluative signals that guide choice (Montague & Berns, 2002; Montague, King- Casas, & Cohen, 2006; Rangel, Camerer, & Montague, 2008).
Other studies look at more challenging decisions where options have multiple attributes, some of which are positive and some negative (Basten, Biele, Heekeren, & Fiebach, 2010; Hare, Camerer, & Rangel, 2009; Kahnt, Greuschow, Speck, & Haynes, 2011). This more closely resembles deliberative decision-making where each outcome has a novel combination of attributes and the person must sum across various dimensions to arrive at an overall valuation. These studies reliably find activity in the vmPFC region and, moreover, the magnitude of activation in this region strongly correlates with the overall summed values across the multiple attributes (Kahnt et al., 2011). This suggests affective mechanisms in vmPFC, which are known to operate during intuitive guidance of action, continue to provide the key evaluative signals during slower, more deliberative decisions.
Additional evidence for affective redeployment comes from studies examining episodic future simulation (Gerlach, Spreng, Madore, & Schacter, 2014; Johnson, Nolen-Hoeksema, Mitchell, & Levin, 2009; Johnson et al., 2006; Mitchell et al., 2009; Spreng & Schacter, 2012). Researchers instructed participants to either construct detailed episodic prospections that involve the person's own goals or prospections that are not goal-relevant. There are, of course, intimate connections between one's goals and evaluative processing. When envisioning a goal-relevant scenario, it is expected that evaluative processes will activate, while a scenario that is not goal-relevant will not elicit strong evaluations. These studies find that vmPFC valuation regions, the regions that subserve intuitive affective processing, are activated during goal- relevant episodes (see Stawarczyk & D'Argembeau, 2015 for a recent meta-analysis). This finding supports the idea that during deliberation when prospective episodes are called to mind, affective systems are deployed to provide evaluative information about these episodes.
The massive reuse view provides an intriguing view of the relationship between intuitive guidance of action and deliberation. The model says we do not have two separate processors operating by very different principles. Rather, intuitive guidance and deliberation are fundamentally connected. There is extensive evidence from psychology and neuroscience that supports the model. First, there is clear evidence that humans have the mental machinery to construct sensorily rich episodic prospections. Second, there is evidence from neurobiology, neuropsychology, and neuroimaging that affective systems are redeployed during deliberation, and they deliver evaluations of these sensorily rich prospective episodes.
This calls into question the claims by many theorists that deliberation is vastly different from intuition. Deliberation is "verbal," "logical," "rule-based," and "neutral," whereas intuition is just the opposite: "nonverbal," "irrational," "associative," and "emotional." This bifurcated picture—two minds built from wildly different materials—is wrong. Rather, the difference between intuitive guidance and deliberative guidance is much more subtle: It consists in the direction of the "mind's eye." During our ordinary transactions with the environment, intuitive affective systems are tuned to the world and they supply ongoing automatic guidance of action. During deliberation, intuitive affective systems once again take center stage. The only difference is that the "gaze" of these systems is pointed away from the actual situation and instead toward episodic prospective representations constructed in the mind.