The Fred Kavli Plenary Lecture

Terrie E. Moffitt’s expertise is in the areas of lifelong aging, mental health, and longitudinal research methods. She is the associate director of the Dunedin Longitudinal Study, which follows a 1972 birth cohort in New Zealand. She also founded the Environmental Risk Longitudinal Twin Study (E-Risk), which follows a 1994 birth cohort in Britain. Moffitt is a licensed clinical psychologist, with specialization in neuropsychological assessment. Her current service includes chair of the Board on Behavioral, Cognitive and Sensory Sciences at the National Academies of Sciences, chair of the National Institute on Aging’s Data-Monitoring Board for the Health and Retirement Study, and chair of the jury for the Klaus J. Jacobs Prize (Switzerland). She is an elected fellow of the US National Academy of Medicine, British Academy, UK Academy of Medical Sciences, and Association of Psychological Science. Moffitt is recipient of the Stockholm Prize, the Klaus Jacobs Prize and the NARSAD Ruane Prize for her work on mental health, and the Maltilda White Riley Award from the NIH for her recent work on aging processes in midlife adults. Moffitt received her PhD in psychology at the University of Southern California, and completed her postdoctoral training at the UCLA Neuropsychiatric Institute. She enjoys working on her poison-ivy farm in North Carolina. Learn more at www.moffittcaspi.com.

Economics and evolutionary biologists often deal with similar behavioural issues, including the construction of preferences, the relation between normative and descriptive accounts, the significance of violations of normative predictions, and biases for or against uncertainty and risk, to name just a few.  The structure of their research programs, however, differs substantially, and this is particularly salient regarding the justification for normative hypotheses and for the role of optimality. I will address some of these differences, drawing on examples from our experimental behavioural work on decision making across different species, ranging from plants to insects, fish, birds, and mammals. A common thread is the use of violations of optimality predictions to enrich and refine normative (evolutionary) analyses.

The human mind is highly tuned towards narratives or human-interest stories that can justify ongoing actions. The human brain, when confronted with the need to make economic decisions, does not simply maximize a stable utility function as hypothesized by much economic theory. This lecture considers the epidemiology of narratives relevant to economic outcomes, allowing them to “go viral” and spread far, even worldwide, thereby influencing economic outcomes. The 1920-21 Depression, the Great Depression starting in 1929, the so called “Great Recessions” of 1973-75, 1980-82, and 2007-9, the secular stagnation-inequality scare after 2008, and President Donald Trump’s economic revolution, are considered in view of the popular narratives of their respective times. These examples are seen as revealing the importance of the linkage of human brains and now computers through narratives associated with popular models of the economy, and offering new research opportunities for both economics and neuroscience.

One key challenge in memory research is to understand the selectivity of memory consolidation – the process by which memory traces become stabilised over time. A key issue is that some events are remembered, others are not. In recent publications, I have argued that the ‘automatic recording of attended experience’ by which we keep track of daily events is followed by the loss of many but not all memory traces within the day, with only some persisting longer in association with the upregulation of plasticity-related proteins (PRPs) that stabilise synaptic potentiation (cellular consolidation) that is the basis of memory formation. Novelty is adept at upregulating PRPs. Memory traces are encoded in parallel in the cerebral cortex (cortical consolidation), with the additional twist that new information may be more successfully assimilated if it fits well with prior knowledge that has previously been stored in cortex in the form of ‘schemas’.  Prior knowledge is, of course, information with which we are most familiar.  The economic implications of this paradox are not my realm of expertise, but I will conclude with some thoughts about accuracy vs. gist that perhaps each affect our economic behaviour.

Regions of the medial prefrontal cortex are known to function in organizing behavior and emotional decision-making, both key functions disturbed in a range of neurologic and neuropsychiatric disorders. In our laboratory, we are seeking to understand mechanisms underlying these functions by applying optogenetic manipulations and microstimulation to these regions and their corticostriatal circuits. We find that we can interrupt the transition from deliberative decision-making to a habitual mode of decisions to act, and can interrupt habitual and insistently repetitive behaviors. In other experiments, we can selectively disrupt decision-making under different contexts involving weighing the costs and benefits of such choices.

The creation of mechanistic neural models of choice computation can come from bottom-up or top-down. I start with classic topics from the social science of collective behavior. In each of these topics, there are long-standing puzzles that do not appear to be easily resolved by simply observing more behavior. The approach discussed in this talk is therefore unapologetically top-down: We start with behavioral observation, hypothesize a biological cause for the behavior and investigate that cause at an intermediate level (e.g., neural circuitry), with whatever method is appropriate for understanding the intermediate level. Three examples illustrate this approach: the influence of affiliative hormones (AVP) on risky cooperation; bargaining with private information (behavior and fMRI); and the formation and crash of asset price bubbles (fMRI).

“Why return to Adam Smith?” Because we learn that he had fresh-for-today insights, derived from a modeling perspective that was never part of economic analysis. Smith wrote two classics: The Theory of Moral Sentiments (1759; hereafter Sentiments); and An Inquiry into the Nature and Causes of the Wealth of Nations (1776; hereafter Wealth). In Sentiments it is argued that human sociability in close-knit groups is governed by the “propriety and fitness” of conduct based on sympathy (that surely influenced Darwin). This non-utilitarian model provides new insights into the results of 2-person experimental “trust” and other games that defied the predictions of traditional game theory in the 1980s and 90s, and offers testable new predictions. Moreover, Smith shows how the civil order of “property” grew naturally out of the rules of propriety. Property together with what I call Smith’s Discovery Axiom then enabled his break-through in Wealth that defined the liberal intellectual and practical foundation of two centuries of Western economic growth.

A hallmark of decision-making in primates is contextual sensitivity: a given stimulus can lead to different decisions depending on the context in which it is presented. This kind of flexible decision-making depends critically upon gating and integration of context-appropriate information sources within the brain. We have analyzed neural mechanisms underlying gating and integration in animals trained to perform a context-sensitive decision task. Surprisingly, both relevant and irrelevant sensory signals are present within frontal lobe circuits that form decisions, implying that gating occurs very late in the process. Dynamical systems analysis of the neural data, combined with a dynamical recurrent network model, suggests a novel mechanism by which gating and integration are combined in a single process.

A reflection on recent advances and questions on the neuroscience of feelings, with an emphasis on the different neural platforms required for sensing affective states.

We investigated basic neuronal reward and risk processes important for decision making using neurophysiological methods in monkeys and brain imaging in humans. Informed decisions between different rewards are based on predictions about future outcomes. We investigated the nature of reward predictive neuronal signals in the amygdala by manipulating the informative nature of the predictive stimulus. We changed the contextual background reward while keeping stimulus reward constant. True reward predictive responses reflected the difference between background and stimulus reward, suggesting that reward contingency rather than simple stimulus-reward pairing (contiguity) determined the predictive neuronal responses. Reward value appears to depend on the individual decision maker and the environment, and hence is subjective. Dopamine neurons in monkeys, and likely downstream striatal activations in humans, discounted reward value across temporal delays of a few seconds despite unchanged objective reward value, suggesting subjective value coding. Reward predictions inform about probability distributions of reward values with varying degrees of risk. Subpopulations of orbitofrontal and striatal neurons, and most dopamine neurons, showed adaptation of reward related responses to the mean and variance (risk) of predicted probability distributions of reward value. These data suggest matching of distributions between neuronal output responses and reward input, resulting in effective reward coding akin to sensory adaptation. Adaptive teaching signals provide stability of learning and established performance in noisy environments. Adaptive neuronal coding may explain such behavioural phenomena as reference dependent coding. The processing of risky outcomes depends on the subjective perception of risk and, separately, on the personal attitudes of individual decision makers towards risk. Dopamine and orbitofrontal neurons in monkeys showed distinct risk signals that were unlikely to constitute value or utility signals. In humans, risk signals and the influence of risk on value signals covaried with individual risk attitudes in subregions of prefrontal cortex, suggesting subjective coding of risk and its influence on reward value. Taken together, these data demonstrate the nature of neuronal reward predictions and suggest subjective coding of key reward variables via temporal delays and adaptive processes in main reward structures of the brain.