How can we continue to analyse human decision making in an age when free will is considered to be an illusion? The answer lies in a behavioral approach. Instead of thinking of decisions as the result of rational human evaluation, we would benefit from discussing them as conditional behavioral responses. This allows us to envision many different algorithmic agents as decision makers and challenges different intellectual traditions to compare notes on where this line of thinking leads.
What do shoppers buying groceries, plants growing their roots in the soil, computer programs shifting the flow among blocks of code, and nations choosing leaders through elections have in common? They are all making decisions. Whether this seems obvious or startling of course depends on how one thinks about decision making.
Researchers at the University of Kentucky from a diverse array of disciplines—computer science, agriculture, political science, biology, linguistics, business, medicine, economics, psychology, and education—recently discovered that they all work on decision making and might be able to trade ideas to good effect. This soon led to an interdisciplinary seminar series and new approaches for training graduate students. In the spring of 2011 came an international conference with funding from the US National Science Foundation and the UK Provost to see what happened when some leading researchers addressed this broad perspective. The conference spawned a book, Comparative Decision Making, published by Oxford University Press earlier this year, addressing the similarities, differences, and insights generated through this boundary defying reach. In this article, we will sketch some of these findings and emphasize the potential for addressing decision making in a comparative way. We hope this will whet your appetite for the book itself, where these ideas are elaborated in greater depth and completeness.[ms-protect-content id=”9932″]
Two decision making concepts
Aristotle’s idea that a decision is a choice among options has long dominated the humanities and the social sciences. A rational human decision maker consciously identifies the options, establishes the value or utility of each, and then selects the best. Personal utilities, the wild card in this formulation, can be inherited, culturally entrained, learned, or may simply reflect physiology or emotion. The concept of free will, major grist for the philosopher’s mill, is associated with this line of thinking about decisions and is the basis for accountability to societal justice systems and to divine judgment in some metaphysical belief systems.
An alternative approach is the behavioral concept of a decision as a conditional response. The decision maker in this case is an “agent”, any entity that can respond to its own condition or to that of its environment or both. Here the conditions trigger responses based on algorithmic processes, as simple as coin flipping or as complex as the neurobiology of human reasoning—and everything in between (such as robots learning to traverse a maze, a band marching in formation, bacteria following a concentration gradient). It is this latter view that opens wide the field for comparative decision making, although the choice process by individual humans is still very much in play.
Philosopher-polymath Bertram Bruce from the University of Illinois plumbs the depths of these two approaches to decision making in the book’s third chapter. He refers to the Aristotelian rational human view as “punctuation”, because it turns on the assumed uniqueness of human cognition, and he refers to the broader conditional response view as “continuity”, because it links a much wider array of phenomena. He argues that both views have much to say about human decision making by joining the traditional emphasis on intent to a more objective emphasis on behavior in environmental and historical context.
These two threads run through several of the book’s chapters, with more emphasis overall placed on the objective “continuity” view that forms the fundamental rationale for the comparative approach.
Moving from foundational topics in decision making based on contemporary microeconomics, philosophy, subjectivist analysis, and game theory, we address the implications of new findings in human neurobiology. Next comes a diverse array of perspectives linked by the insights they provide into how decision making can go “wrong”. The book’s final two chapters explicitly consider the emerging field of decision support: approaches taken to improving the way that decisions get made in human society. In the remainder of this section, we briefly sketch four of the chapters to indicate some of the insights arising from this diversity of perspectives and approaches, with emphasis on problematic decision making.
In a cross-species analysis, psychologists Edmund Fantino and Stephanie Stolarz-Fantino (University of California San Diego) compare the tendencies of humans and pigeons to fall victim to several common errors of decision making. The first of these is “base rate neglect”, in which a decision maker may undervalue information about outcomes in the general population in favor of case-specific information (see the taxicab problem in Figure 1). By challenging pigeons trained to choose by pecking response keys and humans recording their choices to analogous options, the researchers found that pigeons are often better able than humans to balance short-term experiences against general expectations and thus can be less susceptible to base-rate neglect. We humans are quite susceptible to this error, for example in interpreting the results of medical tests or the testimony of eyewitnesses (see Figure 1). It appears that response patterns learned from related tasks can be responsible for this error: “educated” pigeons can become as susceptible as humans to this bias! (Solution to the taxicab problem: 2/3.)
Fantino and Stolarz-Fantino also compared the ability of humans and pigeons to properly ignore “sunk costs”, unrecoverable investments that cannot influence the benefits of future decisions (see the sunk-cost example in Figure 2). Although both species were susceptible to influence by sunk costs, both were able to avoid serious blunders when the expense of essentially “throwing good money after bad” was high enough and the means of escaping this trap was clear. Other human examples of the sunk cost effect include the “Concorde fallacy” and the rationale for continued American involvement in the Vietnam war. (Solution to the vacation conundrum: go to Wisconsin!) In sum, then, decision makers are susceptible to logical errors, and humans are not always superior in avoiding them.
In a similar vein, Ifat Levy at the Yale University School of Medicine studies human preferences between making risky decisions (in which the probability of each of two outcomes is known) and making ambiguous decisions (in which the probabilities of the two outcomes is unknown). Humans strongly prefer risky decisions to ambiguous ones, even in the absence of any clear justification. Yet despite marked differences in behavior under risk and ambiguity, brain scans indicate that the same brain systems are involved in both. Innovative research with humans using brain scans is clarifying the neural mechanisms involved in decision making, opening the way for promising new syntheses like neuroeconomics.
Here we turn to a rather different way of understanding some of the shortcomings of human decision making by revisiting the theme of human rationality. Hugo Mercier (CNRS national research center, Lyon, France) argues that reasoning arose less as a means of solving problems per se and more as a system for building and supporting an argument: the argumentative theory of reasoning. He emphasizes that the hyper-communicative human social environment and benefits of social dominance drove the development of “motivated reasoning”, causing it to overshadow objective reasoning. This intriguing line of thinking has the potential to explain our tendency toward a confirmation bias, which is interpreting evidence with partiality toward our current beliefs—simply post-hoc rationalizing in extreme cases. The dominance of argumentative reasoning may thus indirectly account for some of our inadequacies as logical decision makers, resulting in susceptibility to fallacies like those addressed by Fantino and Stolarz-Fantino. Although Mercier marshals a diverse array of evidence in support of this view, he emphasizes the importance of further scrutiny and the generation of additional cross-disciplinary dialogue, in accord with the comparative approach to decision making put forward in the book.
Shifting gears again, we move to decisions about guilt and innocence in the US criminal justice system from the perspective of attorneys Sarah Crowley (Gross Belsky Alonso LLP, San Francisco) and Peter Neufeld (The Innocence Project, New York, NY). This story begins with the 289 wrongfully convicted people (now 303, as of March, 2013) who have been exonerated through belated DNA analysis of bodily fluids obtained at crime scenes. (Note: These data are from the New York Innocence Project, which has now spawned many other successful projects in the US and other countries.) These exonerations expose a serious systemic accuracy problem in criminal judgments, which are attributed to an overemphasis on “finality”. Here the decision maker is properly considered to be the criminal justice system rather than any single individual, operating under a considerable amount of societal pressure. As in many personal or institutional situations, the decision can be made sooner, providing closure, based on information at hand but at greater risk of error—or the decision can be made more deliberately, potentially based on more information but delaying a resolution. Recognizing this trade-off and making sure the system reflects societal consensus on how it should be resolved are essential goals. Crowley and Neufeld indicate a number of steps that should be taken to improve decision making in this context.
What can comparative decision making accomplish?
Faced with the diversity of decision making in these examples and in many others that fit the concepts presented at the beginning of this article, what can the comparative approach have to offer? Here are some of the more important potential benefits:
Broad educational training: As students begin to make career choices, they may find themselves pigeon-holed into ever-narrower conceptual frameworks and methodologies, to the detriment of their ability to adjust to a rapidly changing set of problems that need solving. It can be a revelation that other fields have their own powerful approaches and insights into the meaning and methods of decision making.
Research advancement through cross-fertilization: Similarly, if the excitement level from our own comparative efforts is any indication, researchers in the many fields that address decision making can benefit from collaborating across traditional fields of enquiry. Although academic and other institutions have an unfortunate tendency to pay lip service to interdisciplinary initiatives, there is much potential for comparative decision making to emerge from the bottom (interactions among individuals) up (into the organizational hierarchy).
Decision support: Individual human and institutional decision makers often need substantive or procedural support in reaching the most effective decisions. Both the decision makers and decision support personnel would benefit from the comparative perspective in gaining access to a full range of options, tools, and creative thinking. In particular, powerful computational methods can and are being developed to advance this cause (the book’s chapter 14 provides such an example).
Linkage of science and the humanities: This has become an ever more elusive goal, especially with contemporary science rushing off in many directions at once, but the topic of decision making certainly provides a potential bridge. The punctuated and continuity decision making concepts described by Bruce above represent both the dichotomy and the opportunities to bridge it. We need more explicit attempts to consider decision making from both perspectives at once, such as a historical analysis of human mate choice or the study of musical preference (also, see the next heading).
Insights into justice, morality, and social norms: Decisions are at the heart of these topics, and there is a large potential comparative dimension across nations, cultures, and even species. The biological origins of morality are under study in our own species, in extinct human species (based on archaeological clues), and in many other organisms that cooperate and are altruistic, particularly toward kin.
Human uniqueness: Humans can only be unique by comparison, and decision making has traditionally been acknowledged as one of the ways we stand apart (as in the Aristotelian view). But many more analyses like those of Fantino and Stolarz-Fantino may be needed to fully establish the differences in degree and in kind, if any, between human decision making and that of other organisms and entities. What we can claim for humans alone may be a shrinking landscape as these comparisons continue.
Some take-home messages
The potential accomplishments indicated above suggest considerable scope for the further development of comparative decision making. We welcome comments and suggestions on these ideas. We close with the “messages” segment taken from the last chapter of the book, summarizing the conclusions we drew from the chapters and from the discussions at the conference:
Humans and other organisms, groups, and some non-living objects and mathematical constructs are capable of behavior we can usefully recognize as decision making. The long intellectual tradition of individual pensive humans as decision makers is alive and well, but the concept of a decision as a conditional response opens the door for the much broader perspective advanced here.
We can often determine optimal, desirable, or acceptable solutions to a particularly diverse array of decision problems based on mathematical analysis. Formal methods generate powerful insights, solve decision problems, and facilitate applications and decision support. Finding optima (best decisions) provides benchmarks against which observed behavior can be compared.
Empirical studies demonstrate that decisions made by humans, other organisms, and groups of organisms often diverge markedly from optima determined from theory. Such mismatches often lead to deeper insights, as recent advances in many areas of decision making analysis have demonstrated.
Many of these non-optimal outcomes may be attributable to decision making features and mechanisms derived from environments strikingly different from those in which decisions must now be made. This environmental shift may cause learned behavior –or evolved behavior –to become inappropriate.
Decision making is often highly constrained, and it is essential to identify and address these constraints in order to understand the outcome and the potential for improvement. Constraints can arise from institutions, natural environment, social environment, physiology, or problem-solving capability.
Similarly, assumptions about key factors in the decision making process (e.g. rationality, scale, self-interest, accuracy) must be continually challenged and adjusted. Information is rapidly accumulating about neural mechanisms, group decision making, selfishness and generosity, institutional shortcomings and other phenomena that alter the way we think about decision making in general and in particular cases.
In important ways, human decision making is quite similar to decision making by other types of agents; but there are many differences of degree (if not kind), some of which are large. This addresses the important philosophical, biological, and behavioral issue of human uniqueness versus fundamental similarity to other organisms and decision making systems.
We are grateful to all participants at the 2011 Conference on Comparative Decision Making in Lexington, Kentucky, and especially to the chapter authors and commentary writers who contributed with such insight and brio to the book. We acknowledge funding from the US National Science Foundation’s ICES Program (H. Pushkarskya, PI; P.H. Crowley, Co-PI) and the University of Kentucky Provost’s Office. The “take-home messages” and the two figures are material originally published in Comparative Decision Making, edited by P.H. Crowley and T.R. Zentall; these are reproduced here by permission of Oxford University Press (see www.oup.com/us and use the search function).
About the Authors
Philip Crowley (Corresponding author) is an evolutionary ecologist who uses both theoretical and empirical methods to address a wide array of issues in natural systems. He is Professor of Biology at the University of Kentucky, where he has taught and conducted research since 1976. Professor Crowley has served as Director of Graduate Studies in Biology and as Director of the School of Biological Sciences; he received the university’s William B. Sturgill Award for contributions to graduate education in 2012. He is currently an INRA Scholar in Sophia Antipolis, France. Email: [email protected], Tel: +1-859-257-1996 (after 8-1-2013).
Thomas Zentall is a comparative cognitive psychologist who studies the similarities and differences between the behavior of humans and other animals, both the cognitive behavior of other animals and the noncognitive behavior of humans. His areas of study include social learning, concept learning, memory strategies, suboptimal gambling behavior, and the effect of effort on the value of the reward that follows. He is Professor of Psychology at the University of Kentucky.