The present projects of our group adopt a model-based and interdisciplinary perspective to study the role of emotional and social states for decision-making. This is a promising research area since there is increasing evidence not only for the importance of such states for economic decision-making but also for the possible contributions of studying them with the help of neuroscientific tools (Camerer et al., 2005; Glimcher and Rustichini, 2004, Rangel et al. 2008). The focus of this research project is to understand the neural mechanisms underlying economic decision-making and how emotions and cognition regulate the underlying decision-making processes. Our project focuses on the study of value-based decision making in (1) an individual and (2) social decision-making context as detailed below.
1) Individual decision making:
Neural Basis of Decision-Making Related Value Signals [PI: Plassmann]
The goal of this project is to better understand the neural basis of subjective value or utility signals that guide economic choices. We are particularly interested in the neural signatures of value signals at the time of choice and consumption and how the brain encodes primary and secondary costs. Project Plan: based on Plassmann’s previous work in this area (Plassmann et al. 2007, 2010; Plassmann et al. 2008, Camus et al. 2009 and Litt et al. 2011) we plan to dissociate neural signatures of different value signals important for human decision-making using model-based fMRI informed by models from behavioral economics, TMS, eye-tracking and MEG. This is a larger-scale project that consists of three related parts.
In a first part, we are investigating how predicted value translates into experienced value and how the brain updates value when the predicted value deviates from the experienced value. For example, imagine that you frequently eat a particular type of cereal bar, but then one day it gives you food poisoning. Does the experience of food poisoning change the predicted value of the same type of cereal bar the next time you consider consuming it? And do changes in predicted value affect experienced value? And how does predicted value change when experienced value and predicted value diverge? Predicted value, experienced value and prediction errors (i.e. deviation of predicted from experienced value) are conceptually different signals, performing different roles in the temporal dynamics of human decision making. In most previous decision neuroscience studies these signals are often correlated and most of the existing literature confounds them (usually by regressing on one of them and assuming that the other one does not exist). This has led to conflicting findings about the role of different brain areas encoding these signals.
In a second part of this project, we are interested in the neural signatures of costs and how costs of different “currencies” get integrated into an overall “net value”. Research from behavioral economics suggests that when purchasing a product, people experience hedonic competition between the anticipated pleasure derived from acquiring and consuming the product and the anticipated losses incurred not only from the money given up in the transaction (product price) and the hassle of executing the payment (transaction cost) but also from pain of paying, the disutility derived from parting with money. It is the trade off between these gains and losses that determines the decision whether or not to purchase (Prelec & Loewenstein, 1998). Previous research in decision neuroscience has investigated how predicted value computations are represented in the brain and whether these representations differ for different modalities such as primary or secondary rewards. However, despite its importance for economic decision-making, little is known about how the human brain computes aversive factors (i.e., costs) during purchasing. This is the central question of this part of the project. In particular, we are interested in whether the representation of costs and their net value integration in the brain differ between monetary costs (e.g. paying money) or somatosensory costs (e.g. tolerating electric shocks) that are matched in economic value.
In a third part of this project, we are interested in dissociating neural correlates of expected value signals from motivational value signals following up on Plassmann’s previous work dissociating expected value and decision saliency signals (Litt et al. 2009). There is evidence from animal and human primate work that we have distinct neural signatures of expected value of a choice option (i.e. “liking”) and how much we are motivated or willing to work for it (i.e. wanting). Interestingly, sometimes these value signals are in conflict. For example, dieters can be in a situation in which they “want” to eat a high caloric food item, but do not like to consume due to negative health consequences. It has proven extremely difficult to translate the idea of this distinction into a valid experimental design. However, we are planning to team up with among others a cognitive psychologist that has developed a reliable task for both of these value signals based on response latency measures. We are planning to use fMRI and pharmacological interventions to dissociate neural underpinnings of expected value signals from motivational value signals.
For parts 1 & 2 of this project we have already conducted pre-tests to adapt materials to a French population, and/ or first pilot data has already been collected. Output & Implications. This line of Plassmann’s research has contributed, and will contribute, to our understanding of how the brain represents value signals that are crucial for an interdisciplinary theory of human decision-making. This basic research is not only fundamentally important for neuroscientists, but also for behavioral decision researchers. This is because we first need to understand more basic questions about which neural systems represent a certain psychological process before we can build on these findings to generate and test new and existing theories of behavioral decision-making.
Cognitive Control of Value Signals [PI: Plassmann]
Plassmann’s previous work on the cognitive regulation of value signals has uncovered which brain mechanisms are involved in up- and down-regulating of options for choice (Hutcherson et al. 2012). Interestingly, we showed that cognitive regulation does not simply lead to increasing or inhibiting neural activity in brain areas that encode preferences, as behavioral measures would suggest. Instead, we found first evidence that cognitive regulation in the form of indulging or up-regulation shifts control to one part of the preference network whereas down-regulation shifts control to a different part of the preference network. These first findings suggest that different computational models might be at play. The goal of the current project is to understand the contribution of each part of the neural preference network and their drivers for cognitive regulation combining model based fMRI and MEG for healthy subject populations and populations with behavioral decision-making disorders (i.e. obese subjects).
Emotional Modulation of Value Signals [PI: Plassmann & Coricelli].
The goal of this line of Plassmann’s research is to elucidate novel psychological mechanisms by which external context factors alter behavioral and neural representations of decision-making related value signals. In these studies, we combine traditional experimental approaches and neuroscientific approaches, drawing on the advantages of each, while overcoming the limits of self-reports.
2) Decision making in dynamic and social settings:
Neuroeconomics of Learning. Regret Learning: its theoretical and neural foundations [PI: Coricelli]
This project aims at improving the understanding of the processes that affect human behavior in the context of interactive repeated choices. We will use both theoretical and neuroscientific methodologies to provide new and more accurate models of learning in interactive settings. We plan to develop a model of adaptive learning where the observation of the outcome of the un-chosen options (counterfactual/fictive outcome) improves the decisions made in the learning process, i.e. regret learning.
Theoretical foundation of regret Learning [PI: Coricelli].
A first difficulty in establishing an analytic foundation for the role of regret is that its effect must be established quantitatively, and qualitatively. Consider for example models of Q-learning. In these models a vector representing the current approximation to the true value is updated in every period by an amount proportional to the prediction error, ignoring the information on the payoff of the other actions. Under some mild technical conditions the process converges to the Q-value that is obtained by following the optimal policy after the first period. It follows that the improvement that can be introduced by considering, through regret, the payoff of the other actions, cannot consist in a better limit behavior, since the optimal one can already be obtained ignoring the payoff of the non chosen actions. An improvement in a different dimension can be introduced: for example, one may show that regret induces a faster convergence to the optimal solution, or a smaller loss in the trajectory leading to the limit. So even if the limit is the same, the speed of convergence to it is faster. A more fundamental problem in any theory of adaptive learning that introduces counterfactual thinking into the analysis of the learning process is what we can call the attribution problem. The problem is easy to understand. The choice of the current action determines the current reward, and also the transition to the next state. Both effects influence the value at the state. Consider the action prescribed by the optimal choice. It may be the case that the reward for that period that can be obtained from a different action is higher. In spite of this, of course the action prescribed by the policy may still be optimal, because the action with the higher current payoff may induce a transition to a “bad" state with low payoffs. A good solution of the problem of integrating regret into adaptive learning when the transition among states depends on the action of the individual is still not available. The fundamental difficulty is in the asymmetry of the information available on the consequences of actions taken: even in the full feedback condition, in which the learner knows the rewards associated with the different actions, he does not know the effect of all the actions on the next state, because this effect is only observable for the action that was really chosen.
Private and social Learning: The role of the Orbitofrontal cortex in emotional learning [PI: Coricelli and Plassmann]
In social learning, the integration of counterfactual thinking into learning is easy from a theoretical point of view, because both effects (on reward and on state transition) of all actions are observed. Project Plan. We will use experimental tasks in which the participant could compare their choices and outcomes with another individual while the two player’s payoffs will remain completely independent. We will directly compare regret/fictive Learning in the two contexts: private and social. The main hypotheses about neural correlates concern the fundamental role of the OFC in regret Learning and its interplay with the mPFC in social Learning.
Interactive decision making and categorization modeling: Learning by doing (sub-cortical) vs. Learning by thinking (cortical) [PI: Coricelli]
With generalization we mean here the set of cognitive mechanisms and rules after which subjects extract from past experience some general knowledge to deal with new, never encountered situations. The current literature reveals a substantial lack of a systematic and robust empirical exploration of this issue.
Reputation building and reputation exploitation: the dynamic interplay between striatal and prefrontal cortex in social learning [PI: Coricelli]
An important question in Learning in games is how beliefs about others’ behavior are generated and exploited in a dynamic setting (i.e. Learning in repeated games). We will investigate the neural basis of reputation and trust during repeated economic games. Our research project aims at understanding the neural correlates of reputation building mechanisms in humans.