Newell & Shanks (N&S) criticize prior work on unconscious influences in decision making for providing insufficient assessment of awareness, leading the authors to question whether unconscious influences should be incorporated as prominent factors in decision theories. While I appreciate their methodological concerns, I am cautious to refute a large body of literature on automatic processes in decision making (Chaiken & Trope Reference Chaiken and Trope1999; Kahneman Reference Kahneman2011; Sloman Reference Sloman1996). I will explore the possible role of unconscious processing in decision making by discussing contextual influences in judgment and choice.
There is a discrepancy between rational decision making, as described by economic theory, and actual choices (Thaler Reference Thaler1980). Both internal and external states (such as visceral factors, framing, and social context) can induce inconsistencies in choice behavior (Cialdini & Goldstein Reference Cialdini and Goldstein2004; Loewenstein Reference Loewenstein1996; Tversky & Kahneman Reference Tversky and Kahneman1981). Danziger et al. (Reference Danziger, Levav and Avnaim-Pesso2011a) report that, prior to food breaks, judges in parole boards less frequently give favorable decisions than after food breaks. The framing effect is manifested, for instance, in the behavior of a majority of consumers who prefer a “75% lean” ground beef product over one having 25% fat, even though there is no difference in the actual product (Levin et al. Reference Levin, Johnson, Russo and Deldin1985). Finally, related to the primes-to-behavior literature reviewed in the target article, descriptive social norms are so powerful in directing behavior that people are even willing to increase their own energy consumption to match the consumption level of their neighbors (Schultz et al. Reference Schultz, Nolan, Cialdini, Goldstein and Griskevicius2007).
These examples raise many questions about the awareness and intentionality of the decision maker. Why is consumer preference affected by the positive or negative presentation of a piece of information? Why would one use more energy than needed – and pay for it – just because others use a lot of energy? And can judges sleep at night peacefully knowing that someone else is behind bars because they were hungry when they made their parole decision? Economically, these choice biases do not make sense, and based on the discussion following the publication by Danziger et al. (Reference Danziger, Levav and Avnaim-Pesso2011a), the legal community objects to the idea that meal breaks influence judicial decisions (Danziger et al. Reference Danziger, Levav and Avnaim-Pesso2011b; Weinshall-Margel & Shapard Reference Weinshall-Margel and Shapard2011).
Recent neuroscience literature has shed light on the underlying mechanisms of framing effects in situations where subjects choose between a positively or negatively framed risky lottery versus a sure outcome. This research suggests that framing effects are mediated by emotional brain areas (amygdala), whereas resisting these effects co-occurs with activation in the anterior cingulate cortex (ACC), a brain region related to conflict detection (De Martino et al. Reference De Martino, Kumaran, Seymour and Dolan2006). These findings are consistent with the expectations of dual-process theories, as they suggest an interplay between initial emotional reactions (System 1) and suppressing control processes (System 2) in the formation and resistance of framing effects, respectively (Kahneman & Frederick Reference Kahneman and Frederick2007). Two further experiments have strengthened these claims. First, individuals with a certain gene variant have a stronger coupling between the ACC and amygdala and are able to resist framing effects better than other individuals (Roiser et al. Reference Roiser, De Martino, Tan, Kumaran, Seymour, Wood and Dolan2009). Second, people with autism spectrum disorder do not show the same pattern of emotional (skin conductance) responses to positive and negative frames compared with control subjects; they also exhibit weaker susceptibility to framing effects (De Martino et al. Reference De Martino, Harrison, Knafo, Bird and Dolan2008). Taken together, this research indicates that largely inborn characteristics can influence the strength of framing effects.
The tendency to follow the behavior of others has been proposed to be driven by error detection and subsequent adjustment (Montague & Lohrenz Reference Montague and Lohrenz2007). Klucharev et al. (Reference Klucharev, Hytönen, Rijpkema, Smidts and Fernandez2009) tested this hypothesis with functional magnetic resonance imaging in the context of facial attractiveness estimation and found that deviations from a stated general group opinion was associated with activation in brain regions that also activate during erroneous responses in simple trial-and-error tasks. The strength of the “error response” was indicative of the subsequent behavioral adjustment toward the group opinion. Intentional adaptation of the reported attractiveness ratings is highly unlikely for two reasons. First, due to extensive requirements for memory of faces and attractiveness ratings, and second, because the behavioral adaptation is also reflected in the neural representation of the stimulus value, suggesting a true modification of opinion beyond mere social gratification (Zaki et al. Reference Zaki, Schirmer and Mitchell2011). Social context also modifies the activation of the reward network for targets other than faces and adjusts the neuronal representation of long-term memories (Campbell-Meiklejohn et al. Reference Campbell-Meiklejohn, Bach, Roepstorff, Dolan and Frith2010; Edelson et al. Reference Edelson, Sharot, Dolan and Dudai2011). Together, these findings suggest that social influence in decision making is mediated by adapted value estimates and memories.
The literature reviewed here unfortunately cannot give conclusive information about the presence or absence of unconscious influence in the framing effect and social conformity, because the experimental procedures did not include rigorous assessments of awareness. Regardless, the gained knowledge gives many weak indicators for unintentionality, if not unawareness. First, the neuroscientific findings of framing effects comply with the expectations of dual-process theories and show that inborn features may influence the strength of behavioral framing effects, indicating that the decision process is systematically different between groups of people in a simple and reasonably neutral decision task. Second, even a single exposure to a descriptive social norm can modify the value of an item possibly through basic and automatic reinforcement learning mechanisms. One noteworthy aspect is that the dual-process accounts do not necessitate that System 1 influence is uncontrollable in a top-down fashion (Chaiken & Trope Reference Chaiken and Trope1999). While in some conditions there might be an unconscious effect, in other situations the influence of framing or social norms can be intentionally acknowledged and controlled by the decision maker.
A logical next step is to conduct further tests that measure a decision maker's awareness of the effect of framing and social norms. N&S give many good pointers for designing methodologically sound experiments, but one should be careful not to influence the decision-making process with the awareness assessment. Highlighting aspects of the decision-making task can change the course of the decision-making process by increasing attention and top-down control.
Newell & Shanks (N&S) criticize prior work on unconscious influences in decision making for providing insufficient assessment of awareness, leading the authors to question whether unconscious influences should be incorporated as prominent factors in decision theories. While I appreciate their methodological concerns, I am cautious to refute a large body of literature on automatic processes in decision making (Chaiken & Trope Reference Chaiken and Trope1999; Kahneman Reference Kahneman2011; Sloman Reference Sloman1996). I will explore the possible role of unconscious processing in decision making by discussing contextual influences in judgment and choice.
There is a discrepancy between rational decision making, as described by economic theory, and actual choices (Thaler Reference Thaler1980). Both internal and external states (such as visceral factors, framing, and social context) can induce inconsistencies in choice behavior (Cialdini & Goldstein Reference Cialdini and Goldstein2004; Loewenstein Reference Loewenstein1996; Tversky & Kahneman Reference Tversky and Kahneman1981). Danziger et al. (Reference Danziger, Levav and Avnaim-Pesso2011a) report that, prior to food breaks, judges in parole boards less frequently give favorable decisions than after food breaks. The framing effect is manifested, for instance, in the behavior of a majority of consumers who prefer a “75% lean” ground beef product over one having 25% fat, even though there is no difference in the actual product (Levin et al. Reference Levin, Johnson, Russo and Deldin1985). Finally, related to the primes-to-behavior literature reviewed in the target article, descriptive social norms are so powerful in directing behavior that people are even willing to increase their own energy consumption to match the consumption level of their neighbors (Schultz et al. Reference Schultz, Nolan, Cialdini, Goldstein and Griskevicius2007).
These examples raise many questions about the awareness and intentionality of the decision maker. Why is consumer preference affected by the positive or negative presentation of a piece of information? Why would one use more energy than needed – and pay for it – just because others use a lot of energy? And can judges sleep at night peacefully knowing that someone else is behind bars because they were hungry when they made their parole decision? Economically, these choice biases do not make sense, and based on the discussion following the publication by Danziger et al. (Reference Danziger, Levav and Avnaim-Pesso2011a), the legal community objects to the idea that meal breaks influence judicial decisions (Danziger et al. Reference Danziger, Levav and Avnaim-Pesso2011b; Weinshall-Margel & Shapard Reference Weinshall-Margel and Shapard2011).
Recent neuroscience literature has shed light on the underlying mechanisms of framing effects in situations where subjects choose between a positively or negatively framed risky lottery versus a sure outcome. This research suggests that framing effects are mediated by emotional brain areas (amygdala), whereas resisting these effects co-occurs with activation in the anterior cingulate cortex (ACC), a brain region related to conflict detection (De Martino et al. Reference De Martino, Kumaran, Seymour and Dolan2006). These findings are consistent with the expectations of dual-process theories, as they suggest an interplay between initial emotional reactions (System 1) and suppressing control processes (System 2) in the formation and resistance of framing effects, respectively (Kahneman & Frederick Reference Kahneman and Frederick2007). Two further experiments have strengthened these claims. First, individuals with a certain gene variant have a stronger coupling between the ACC and amygdala and are able to resist framing effects better than other individuals (Roiser et al. Reference Roiser, De Martino, Tan, Kumaran, Seymour, Wood and Dolan2009). Second, people with autism spectrum disorder do not show the same pattern of emotional (skin conductance) responses to positive and negative frames compared with control subjects; they also exhibit weaker susceptibility to framing effects (De Martino et al. Reference De Martino, Harrison, Knafo, Bird and Dolan2008). Taken together, this research indicates that largely inborn characteristics can influence the strength of framing effects.
The tendency to follow the behavior of others has been proposed to be driven by error detection and subsequent adjustment (Montague & Lohrenz Reference Montague and Lohrenz2007). Klucharev et al. (Reference Klucharev, Hytönen, Rijpkema, Smidts and Fernandez2009) tested this hypothesis with functional magnetic resonance imaging in the context of facial attractiveness estimation and found that deviations from a stated general group opinion was associated with activation in brain regions that also activate during erroneous responses in simple trial-and-error tasks. The strength of the “error response” was indicative of the subsequent behavioral adjustment toward the group opinion. Intentional adaptation of the reported attractiveness ratings is highly unlikely for two reasons. First, due to extensive requirements for memory of faces and attractiveness ratings, and second, because the behavioral adaptation is also reflected in the neural representation of the stimulus value, suggesting a true modification of opinion beyond mere social gratification (Zaki et al. Reference Zaki, Schirmer and Mitchell2011). Social context also modifies the activation of the reward network for targets other than faces and adjusts the neuronal representation of long-term memories (Campbell-Meiklejohn et al. Reference Campbell-Meiklejohn, Bach, Roepstorff, Dolan and Frith2010; Edelson et al. Reference Edelson, Sharot, Dolan and Dudai2011). Together, these findings suggest that social influence in decision making is mediated by adapted value estimates and memories.
The literature reviewed here unfortunately cannot give conclusive information about the presence or absence of unconscious influence in the framing effect and social conformity, because the experimental procedures did not include rigorous assessments of awareness. Regardless, the gained knowledge gives many weak indicators for unintentionality, if not unawareness. First, the neuroscientific findings of framing effects comply with the expectations of dual-process theories and show that inborn features may influence the strength of behavioral framing effects, indicating that the decision process is systematically different between groups of people in a simple and reasonably neutral decision task. Second, even a single exposure to a descriptive social norm can modify the value of an item possibly through basic and automatic reinforcement learning mechanisms. One noteworthy aspect is that the dual-process accounts do not necessitate that System 1 influence is uncontrollable in a top-down fashion (Chaiken & Trope Reference Chaiken and Trope1999). While in some conditions there might be an unconscious effect, in other situations the influence of framing or social norms can be intentionally acknowledged and controlled by the decision maker.
A logical next step is to conduct further tests that measure a decision maker's awareness of the effect of framing and social norms. N&S give many good pointers for designing methodologically sound experiments, but one should be careful not to influence the decision-making process with the awareness assessment. Highlighting aspects of the decision-making task can change the course of the decision-making process by increasing attention and top-down control.
ACKNOWLEDGMENT
Supported by the aivoAALTO project of the Aalto University.