Understanding the Hows and Whys of Decision-Making: From Expected Utility to Divisive Normalization

Abstract

Over the course of the last century, economists and ethologists have built detailed models from first principles of how humans and animals should make decisions. Over the course of the last few decades, psychologists and behavioral economists have gathered a wealth of data at variance with the predictions of these economic models. This has led to the development of highly descriptive models that can often predict what choices people or animals will make but without offering any insight into why people make the choices that they do-especially when those choices reduce a decision-maker's well-being. Over the course of the last two decades, neurobiologists working with economists and psychologists have begun to use our growing understanding of how the nervous system works to develop new models of how the nervous system makes decisions. The result, a growing revolution at the interdisciplinary border of neuroscience, psychology, and economics, is a new field called Neuroeconomics. Emerging neuroeconomic models stand to revolutionize our understanding of human and animal choice behavior by combining fundamental properties of neurobiological representation with decision-theoretic analyses. In this overview, one class of these models, based on the widely observed neural computation known as divisive normalization, is presented in detail. The work demonstrates not only that a discrete class of computation widely observed in the nervous system is fundamentally ubiquitous, but how that computation shapes behaviors ranging from visual perception to financial decision-making. It also offers the hope of reconciling economic analysis of what choices we should make with psychological observations of the choices we actually do make.The postenlightenment study of human decisionmaking has its roots in the work of mathematicians and economists like Blaise Pascal, Daniel Bernoulli, Adam Smith, and David Ricardo. These scholars sought to explain regularities in human behavior from first principles. For these early theorists, that meant defining algorithms that maximized individual gain, profit, or well-being and that thus gave rise to observable market level regularities. These scholars took as their focus the study of "why"-why it was that people made the choices they did from a mathematical and theoretical point of view. While revising many of the key assumptions of these early scholars, economists of the early 1900s like Paul Samuelson (Samuelson 1947) and John von Neumann (Neumann and Morgenstern 1944) made huge strides in this direction by defining how decision-making ought to work in humans. They developed theories that fundamentally described how "optimal" decision-makers should behave.In the 1960s, however, it started to become clear that humans did not always behave in exactly the ways predicted by these theories. Work by economists like Maurice Allais (Allais 1953) and psychologists like Amos Tversky and Daniel Kahneman (Kahneman and Tversky 1979) began to reveal systematic patterns of behavior that violated the prediction...

Segregation, connectivity, and gradients of deactivation in neural correlates of evidence in social decision making

“…Consistently with these findings, functional imaging studies of decision making in man have retrieved sparse cortical signals from the regression of the signal on the difference between the decision values of the chosen and non-chosen options (Heekeren et al 2008). It has also been argued that this mechanism may also apply to preference-based choice (Krajbich and Rangel 2011;Cisek 2012;Glimcher 2015), and to choices in social cognition (Shadlen and Kiani 2013), where evidence in this respect remains scant (Ruff and Fehr 2014).…”

Predictive regulation and human design

“…Thus, when a behavior delivers a result better than predicted, certain neurons reward the 'choosing circuits' by secreting a pulse of dopamine. This encourages the brain to store the memory and repeat the behavior ( Glimcher, 2014 ; Schultz, 2015 ). This type of reward learning is mathematically optimal, and since it is present in the fruit fly, it was probably already present in the brain of our last common ancestor the urbilaterian worm.…”

Rational Imprecision: Information-Processing, Neural, and Choice-Rule Perspectives

“…This appendix is intended to supply the interested reader with a brief overview of the origins and ubiquity of the normalization forms in the neuroscientific literature. A more detailed review can be found in Glimcher (2014).…”