1. ECONOMICS AS A NATURAL SCIENCE

Underlying my reasoning is the understanding of economics as a natural science, conceptually analogous to physics or biology. In this view, the object of economics is choice behaviour, in its diverse and multifaceted complexity. A few comments on this stance are in order.

First, while differing from classic definitions based on optimality (Robbins Reference Robbins1935, Samuelson Reference Samuelson1970), the understanding of economics as a natural science does not necessarily reduce it to a purely descriptive (as opposed to normative) science. Indeed, the prescriptive/normative endeavour – devising and describing the best possible set of choices given preferences and constrains – certainly is part of the analysis of choice behaviour, whether or not actual choosers abide by those prescriptions. Let me illustrate this point with an example. Like any natural science, economics describes facts of life. One particular fact of life is that exponentials are the only discounting functions that ensure intertemporal consistency (which, per se, is certainly desirable). Another fact of life is that in many circumstances people's discounting functions are better described as hyperbolic. Both these facts of life pertain to choice behaviour and are thus legitimately described by economics.

Second, defining economics as the science of choice behaviour does not imply taking a position on the question of interest here – whether neuroscience can contribute to economics. Indeed, a full and satisfactory theory of choice behaviour could in principle be formulated ignoring completely the psychological and physiological processes that lead to choice. Alternatively, psychological and physiological evidence could be an essential part of the theory. Both these possibilities are left open by the definition.

Third, in the science of choice behaviour, like in any other natural science, different and competing theories should be distinguished empirically based on the usual criteria of accuracy, generality and parsimony. Of two alternative theories, we should deem preferable the one that best satisfies these criteria.

2. TERMS OF THE SYLLOGISM

Although the term “neuroeconomics” might suggest otherwise, that between economics, psychology and neuroscience truly is a ménage à trois. Loosely put, these three natural sciences have different objects. Economics is the science of choice behaviour, psychology is the science of thought, and neuroscience is the science of the brain. These rough definitions, given independently of one another, imply that economics, psychology and neuroscience could in principle be separate sciences. The alternative hypothesis is that the three sciences, while operating at different levels, share a fundamental unity. In this latter view, methods, empirical results and theories proper of one science may bestow and support better theories in another science. Discussing whether neuroscience can contribute to economics essentially amounts to addressing whether there is fundamental unity or fundamental separateness between these three sciences.

My argument for unity is based on the following syllogism (of sorts). If neuroscience will contribute to economics, it will do so by the way of psychology. In other words, discoveries about the brain will hopefully lead to better theories of thought, which in turn will hopefully lead to better theories of choice behaviour. The syllogism is based on two premises – that neuroscience can contribute to psychology, and that psychology can contribute to economics. These premises, I will argue, are well documented already. It thus follows that neuroscience can contribute to economics. Notably, this proposition is a statement of possibility. Whether any discovery about the brain will in fact lead to better economic theories is ultimately an empirical question. Current examples might not be the most compelling, but the field is in its infancy and time will tell.

In the following, I summarize the arguments that support the two premises of the syllogism.

3. PSYCHOLOGY AND ECONOMICS

Early economists were concerned as much with moral sentiments as they were with the wealth of nations.Footnote ¹ Utilitarian theories of economic choice, deeply rooted in psychological theories of pain and pleasure, were built on a strong concept of value. Over generations, however, economists gradually embraced weaker assumptions. In the 1930s, it was finally realized that economic theory could dispense of subjective value all together, and simply be founded on revealed preferences (Samuelson Reference Samuelson1938). For the non-economist, it is worth emphasizing that the formulation in terms of revealed preferences was a scientific and mathematical triumph, for it freed economic theory from the burden of scrutinizing the chooser's mind. In this light, the reluctance of mainstream economics to re-open that (finally sealed) Pandora's box appears very understandable.Footnote ² Yet, current challenges to the standard model are hard to ignore.

While revisiting the relationship between economics and psychology, a few remarks are in order. First, although neoclassical economics is arguably “mindless” (Gul and Pesendorfer Reference Gul, Pesendorfer, Caplin and Schotter2008), it is not quite “psyche-less”. Indeed, the revealed preferences approach did not quite remove psychology from economic theory. Rather, neoclassic economics embraced one particular psychological theory – Skinnerian behaviourism.Footnote ³ The central tenet of behaviourism is that all behaviours result from stimulus-response associations learned through experience. In other words, behaviourism describes the psyche as a large look-up table where different sets of stimuli are associated to different responses. In this view, mental states and cognitive processes either do not exist or do not need to be investigated by psychology. In the case of choice behaviour, subjects learn to associate a set of options to one particular action. Thus after training, “choosing” simply amounts to retrieving the proper association.Footnote ⁴ Behaviourism thus denies the existence of any other mental process, including the assignment of value. In other words, according to behaviourism value is not a psychologically real entity; it is merely a construct used by the observer to describe choices. Subsumed in economics, behaviourism thus translates into revealed preferences.

Importantly, although it plays a foundational (if largely implicit) role in neoclassical economics, behaviourism was and remains a psychological theory. In this respect, it should be noted that, as a psychological theory, behaviourism is largely outdated. Starting with Chomsky (Reference Chomsky1959), cognitive scientists indeed recognized that behaviourism fails the scientific standard of Einstein's initial quote. In nearly every domain of psychology, from language to perception, the associative model proved exceedingly simplistic and better psychological theories were formulated in terms of mental states and cognitive processes.Footnote ⁵ With respect to choice behaviour, for example, it was proposed that choosing entails two distinct mental processes – the assignment of values to the available options, and the comparison (or decision) between these values.Footnote ⁶ The continuing prominence of the revealed preferences approach in mainstream economics can thus be viewed as somewhat paradoxical. Behaviourism, a psychological theory, is essentially abandoned within psychology. However, it remains as a cornerstone in another discipline, namely economics.

The relevant question here is whether any cognitive model that provides a better psychological account of choice behaviour than that provided by behaviourism can also lead to better economic theories than those based on revealed preferences. In my view, the answer should be affirmative unless proven otherwise. Indeed, it seems safe to assume that economic theories based on “good” psychological models will be at least as “good as economic theories based on “bad” psychological models. And indeed, numerous current economic models – from prospect theory (Kahneman and Tversky Reference Kahneman and Tversky1979) to models of hyperbolic discounting (Ainslie Reference Ainslie1992) – are rooted in notions from cognitive science. Hence, although a comprehensive economic theory based on mental processes remains elusive (see below), it seems reasonable to conclude that psychology can indeed contribute to economics.

4. NEUROSCIENCE AND PSYCHOLOGY

In a time when MRI scanners bloom in psychology departments around the world, questioning whether neuroscience can contribute to psychology may seem futile. However, MRI scanners and colourful images do not per se address earlier arguments for separateness (von Hayek Reference von Hayek1952, Fodor Reference Fodor1975). In a nutshell, these arguments reject a reductionist claim – that mental states can be precisely mapped into physical brain states and that psychology can thus be ultimately reduced to neuroscience. One challenge to reductionism follows from the observation that identical mental states may be physically instantiated in very different realizations.Footnote ⁷ Another challenge to reductionism comes from the enormous complexity of the brain. In practical terms, realizing that mind-to-brain reduction would amount to devising a large mechanistic model of the brain, where each element of the model represents one neuron and elements in the model are connected in ways that represent the synapses in the brain. That model would faithfully reproduce behaviour, and mental states would then be mapped into states of the model. Unfortunately (or fortunately), the human brain has (conservatively) some 10¹¹ neurons and perhaps 10¹⁴ synapses. Here, the hopeful reductionist may want to take a tip from the statistical physicist.Footnote ⁸ If a microscopic theory of such a large system is impossible when elements are simple and non-interacting particles in a perfect gas, how can it be realized when elements are complex and interacting neurons? This route of reductionism thus seems impracticable.

Nonetheless, I believe that the methods of neuroscience have a lot to offer to psychology. In fact, the distinction between neuroscience and psychology is arguably fictitious. Although the levels of description are certainly different, cognitive processes and brain activity are ultimately one and the same. The problem is not reducing the mind to brain activity – the mind is brain activity. The problem is to understand the mind/brain. In this endeavour, psychological theories and neural models should go hand-in-hand. Of course, neuronal processes are very complex and often hard to decipher, but so are cognitive processes. And in many cases deeper inroads can be made going back and forth between the two levels of description than can be made by working at one level alone.

Examining neuronal responses in a specific condition amounts to studying a particular cognitive process, or some aspect of it. In other words, single neuron recordings and brain imaging are as much methods of psychology as they are methods of neuroscience. This is not just a statement of principles. In many circumstances, different and competing psychological models are hard to disentangle based on behavioural evidence, and measures of brain activity can support or rule out one particular hypothesis.Footnote ⁹ Instances in which neuronal findings have shaped subsequent psychological theories abound. For example, Marr's (Reference Marr1982) model of vision (a psychological model insofar as vision is a mental process) builds on the classic discovery of Hubel and Wiesel (Reference Hubel and Wiesel2005) that neurons in the primary visual cortex respond to oriented bars. Their discovery rules out alternative models of vision, such as models based on punctiform stimuli. Another example is particularly relevant to choice behaviour. Consider a person sitting in a restaurant and choosing between two options on the menu – pizza and salad. While choosing, the person presumably assigns values to the two options and subsequently decides by comparing these values. Within this general framework, two alternative psychological models have been proposed. According to the “good-based model”, values are compared in the space of goods, where a decision is made between pizza and salad. Subsequently, an appropriate action is selected to implement that choice. According to the “action-based model”, values are first attached to different possible actions, and the decision unfolds as a process of action selection. The issue is not fully resolved based on behavioural evidence alone. However, there is consensus that the controversy can be addressed based on neuronal measures.Footnote ¹⁰ In conclusion, neuroscience can certainly contribute to psychology.

5. A HOPEFUL OUTLOOK

The syllogism of neuroeconomics goes as follows. Neuroscientific discoveries can lead to better psychological theories, which can lead to better economic theories; therefore, neuroscience can contribute to economics. Will this actually happen?

Neoclassical economics faces a number of empirical challenges – hyperbolic discounting, loss aversion and framing effects, to mention a few. Current economic models that account for these phenomena maintain the structure of the classic theory, with the addition of one or more psychological parameters that explain deviance from rationality. For example, models of fairness add to the classic utility a term for social preferences (Rabin Reference Rabin1993). Similarly, models of intertemporal choice explain hyperbolic discounting as due to attentional limitations (Benhabib and Bisin Reference Benhabib and Bisin2005), or in terms of multiple selves (Laibson Reference Laibson1997). And models of addiction include the psychological parameter of hedonic value (Bernheim and Rangel Reference Bernheim and Rangel2004). This “patching” approach is undoubtedly a good first step. However, by the standards of a natural science, the patching approach has severe limitations, because new psychological parameters are invoked ad hoc and injected into the theory every time a new fallacy is to be explained. Furthermore, if we imagine collecting the current economic models that describe various deviances form the classic assumptions, the emerging economic agent is rather awkward. She always has stable preferences, except that her choices are also crucially influenced by strict attentional limitations (and nothing else) in a specific set of conditions, by the urgent need of social fairness (and nothing else) in another set of conditions, by the excessive distaste for loss (and nothing else) in yet another set of conditions, and so on. Can we hope for a more comprehensive economic theory?

At the turn of the twentieth century, physics found itself in a situation analogous to that faced by economics today. Newtonian mechanics and electromagnetism, which once seemed to provide a stable and comprehensive system, were challenged theoretically and empirically by phenomena such as black body radiation, the photoelectric effect and the Compton scattering.Footnote ¹¹ Initially, classic theories were patched to account for one effect or the other, but within a few decades a new comprehensive theory emerged. To be sure, quantum physics is grounded on qualitatively different concepts – such as the uncertainty principle and the wave-particle duality – that would have been simply unthinkable in the classic mind frame. Will a “quantum economics” similarly emerge in the coming decades? It is hard to say. But when it does, it may well be grounded on such “unthinkable” concepts as cognitive processes and brain activity.