Current attitudes towards neuroeconomics have been profoundly shaped by the surveys that began to appear in mainstream economic journals around the middle of the last decade. Many economists for example took manifestos such as Camerer et al. (Reference Camerer, Loewenstein and Prelec2005) to be definitive of the entire new field of neuroeconomics – in particular the idea that neuroeconomics amounts to opening up economics to the insights and findings from neuroscience. Camerer et al. (Reference Camerer, Loewenstein and Prelec2005) envisioned that this would eventually revolutionize economic theory. With such revolutionary zeal based on promissory notes rather than well-wrought results, it does not come as a surprise that this version of neuroeconomics, aptly called Behavioral Economics in the Scanner (BES) by Ross (Reference Ross2008), met quite some resistance in the profession. Poldrack (Reference Poldrack2006), Harrison (Reference Harrison2008) and Rubinstein (Reference Rubinstein2008), for example, denounced neuroeconomics for mostly methodological flaws. But the most radical dismissal of neuroeconomics to date is Gul and Pesendorfer (Reference Gul, Pesendorfer, Caplin and Schotter2008). In their case for a ‘mindless economics’, Gul and Pesendorfer observe that on the prevailing revealed preference interpretation, expected utility theory does not make predictions about psychological states or brain states. They conclude from this that psychological states or brain states cannot be invoked to appraise economic theory. Since choice behaviour is the only thing expected utility theory makes predictions about, choice data (or behavioural data) are the only sort of data that can possibly refute expected utility theory.

Camerer et al. (Reference Camerer, Loewenstein and Prelec2005) however was neither the only nor the first manifesto of neuroeconomics. Paul Glimcher's (Reference Glimcher2003) version of neuroeconomics was in many ways altogether different from the version favoured by behavioural economists. Rather than relying on new neuro-imaging techniques such as fMRI and PET scans, which register differential activation of brain areas, Glimcher used single-neuron recordings (single neuron electrophysiology). Furthermore, unlike BES, Glimcher did not look for brain data that could contradict expected utility theory. If anything, Glimcher seemed to be looking for exactly the opposite: speaking as a neurobiologist, he advocated the use of expected utility theory (and game theory) in neuroscience. Indeed, findings as those reported in Platt and Glimcher (Reference Platt and Glimcher1999) and Dorris and Glimcher (Reference Dorris and Glimcher2004) were considered to be indicating that various neurons (especially in area LIP – the lateral intraparietal area of the posterior parietal cortex) encode expected utilities of different choice options. For decades economists had not dared to hope that expected utility theory could be defended on a literal interpretation, as a theory about individual decision-making. But here was Glimcher showing, it seems, that such a defence is viable after all.

This is at least how Glimcher's findings were commonly understood: as supporting rather than undermining expected utility theory (cf. Vromen Reference Vromen2010). But Glimcher's fascinating new book tells a somewhat different story. Glimcher makes clear that what he was looking for in his earlier studies was not a vindication of expected utility theory, but rather a localization of the specific sites in the brain where the last stages of decision-making are implemented. What he found was that the firing rates of neurons in area LIP were linearly correlated with eye movements of monkeys (in so-called saccade tasks). On the basis of these findings and others, Glimcher hypothesizes that LIP is one of the brain areas in which the penultimate stage of all decision-making processes in the brain, the valuation of choice options, occurs. This stage is only followed by the very final stage, that of making choices proper on the basis of the options’ relative values (which occur notably in the fronto-parietal network of brain areas), before motor circuits are activated to produce observable behaviour (such as the eye movements of monkeys). As this final choice mechanism often proceeds in a winner-take-all fashion – the neurons with highest firing rates fully determine what happens next – the ensuing integrated picture of the last two stages of decision-making in the brain is one, Glimcher contends, that very much resembles the ‘arg max’ operation suggested by expected utility theory: first the subjective values of the choice options are computed by neurons, then the option with the highest subjective value is chosen.

The main thesis Glimcher develops and defends in his book is that no matter how different decision-making processes might be related to earlier (‘upstream’) stages, all decision-making processes in the brain eventually terminate in these two final stages: firstly, one of valuation and secondly, one of choice proper. These two stages are physiologically implemented in different brain areas. Whatever happens in these two final stages determines choice behaviour. Given its resemblance with expected utility theory, Glimcher calls this theory ‘Hard-EU’. He contrasts Hard-EU with Soft-EU, which has been popular among economists for decades. Soft-EU entails that preferences can be inferred from (because they are assumed to be revealed by) choices; that agents behave as if they maximize expected utility. Hard-EU goes beyond this in advancing the causal claim that agents behave the way they do because their behaviour is determined by what Glimcher calls expected personal values computed in the brain. This implies that unlike Soft-EU, in which preferences cannot be identified independently of choice behaviour, Hard-EU assumes that expected personal values can be measured independently of choice behaviour.

What Glimcher is arguing here can easily be misunderstood. Glimcher might be taken to argue that neuroeconomists should be in the business of measuring expected personal values, perhaps on a case-by-case basis, in order to predict individual choices. This seems to be suggested by his plea to replace Soft-EU by Hard-EU: positive economic theory should start providing causal-mechanistic explanations of individual behaviour in terms of expected personal values measured in the brain. Rather than inferring preferences indirectly from choice behaviour, preferences can and should now be measured by peering into what is occurring in the relevant areas in the brain. The second thing Glimcher seems to argue is that Hard-EU is at least approximately true. He states, for example, that expected personal values track choice (p. 190). Since Hard-EU explains choice by expected personal values, this seems to entail that Glimcher believes that Hard-EU is true.

Yet Glimcher argues neither that the task of neuroeconomics is to measure expected personal values in the brain so as to explain choice, nor that Hard-EU is approximately true. Starting with the latter, Glimcher does not put forward Hard-EU because he believes it accurately describes decision making in the brain. On the contrary, Glimcher believes that empirical tests will readily reveal that like Soft-EU, Hard-EU is false. Understood as a positive theory about individual choices, expected utility theory, either in its Soft-EU or Hard-EU version, does not have an impressive predictive record. Glimcher recognizes that often choice behaviour does not exhibit the type of internal consistency that axioms of choice theory, such as WARP and GARP, impose on behaviour. Thus there is room for improvement. Glimcher is convinced that in order to eventually arrive at an empirically accurate neuroeconomic theory, we first need to search for neural correlates of the ‘logical objects’ of expected utility theory in situations where expected utility theory does a good job in predicting individual choice. This is what Hard-EU does for us: it forges an inter-theoretic link (or a partial reduction, as Glimcher also calls it) between economic theory and neuroscience. (Glimcher does not believe that all of economics can be reduced to neuroscience – he believes that there are emergent concepts in economics that resist reduction.) As Glimcher puts it succinctly, ‘we begin the neuroeconomic process of theory development with Hard-EU not because expected utility theory is right, but because for a synthetic neuroeconomist, Hard-EU is the right place to start’ (p. 139).

Once the inter-theoretic link between economic theory and neuroscience is forged by Hard-EU, Glimcher argues, we can enter the next phase in the neuroeconomic process of theory development: we can bring what we learn from neuroscience about neural activity in the brain areas relevant for decision-making to bear on further (neuro)economic theorizing. Here we traverse in reverse direction, from neuroscience to economics. Neuroscience is supposed to provide mechanistic constraints on acceptable revisions of economic theory. Glimcher argues for example that we know that the firing rates of neurons encode relative, not absolute values. And we also know that the winner-take-all mechanism does not provide the only way in which choice occurs in the brain. Sometimes choice is determined in a more direct way, as when some activity is produced when the firing rates of corresponding neurons exceed some critical threshold value. Glimcher notes that this is reminiscent of Herbert Simon's ideas about satisficing, and would like such knowledge to have consequences for further economic theorizing. This interplay between high-level theory, provided by economics, and low-level mechanistic constraints, provided by neuroscience, is characteristic of the sort of inter-theoretical cross-fertilization Glimcher is after. Glimcher's project is ambitious: in the end, he tries to accomplish nothing less than an interdisciplinary synthesis of economics, psychology and neuroscience.

Glimcher does not want neuroeconomics to revive Edgeworth's dream of the hedonimeter. Neuroeconomics should not measure the neural correlates of expected utility in the brain so as to predict choice behaviour. Instead, neuroeconomics should try to contribute to the development of an improved positive economic theory, one with greater predictive power than expected utility theory. He puts crisp and clear what neuroeconomics should be ultimately after: ‘Our goal is to produce a successor to Soft-EU that has greater predictive power’ (p. 190). Neuroeconomics’ way to achieve a positive economic theory with greater predictive power is to align economic theory better with the corpus of psychological and neuroscientific thought. The fact that Glimcher occasionally speaks of the need to develop a Behavioural Neuroeconomics signals that he believes that significant changes in economic theory might result from this. However, Glimcher does not believe that the way to go forward in further developing economic theory is to abandon the utility-based approach, as proponents of a heuristics-approach seem to advocate. He is also not in favour of incorporating new parameters in utility-models, as prospect theory aims to do. What we rather need is a revised set of axioms, where the revisions are guided by constraints stemming from psychology and neuroscience. Glimcher does not want to abandon the axiomatic approach in economics, primarily for the reason that the axiomatic approach allows for unambiguous empirical testing of the theory. He sees the joint work he has been doing lately with economists such as Caplin and Dean (Caplin et al. Reference Caplin, Dean, Glimcher and Rutledge2010), following Caplin and Dean's (Reference Caplin and Dean2008) pioneering work, as a first step in that direction.

That this is how Glimcher envisions the future of neuroeconomics is clear to me. What is still not clear is what neuroscience can contribute to the development of a positive economic theory with greater predictive power than expected utility theory. What useful information can knowledge about neural activity provide that is not already entailed in observable choice data? One of the things Glimcher stresses in the book is that deterministic utility models have to give way to random utility models, as developed by McFadden (Reference McFadden2005) and Gul and Pesendorfer (Reference Gul and Pesendorfer2006). We must replace Hard-EU by Hard-REU, Glimcher argues. The pioneers of random utility models did not need insights from neuroscience (or ‘brain data’) to develop their models, however. Choice data, obtained in repeated experiments with lotteries with gradually shifting prizes, sufficed to motivate the development of random utility models. Gulf and Pesendorfer's (2008) misgiving seems to hit the target here: bringing in brain data does not seem to add anything to what we already knew from looking at choice data. One of the cornerstones of Glimcher's neuroeconomics, that expected subjective values in the brain always track choice, only seems to reinforce Gul and Pesendorfer's view that choice data are all economists need to tell whether or not their models are empirically accurate.

Glimcher might retort here that we can learn from neuroscience that it is due to the inherent stochasticity in neural activity in relevant brain areas such as LIP that choice behaviour tends to be indeterministic. This surely deepens our understanding of why choice behaviour happens to be indeterministic. But do we need this deeper understanding to improve our economic theories? When our sole aim is to develop an economic theory with greater predictive power than expected utility theory, just having the relevant choice data seem to suffice. It might be argued, perhaps, that knowing the underlying neural mechanisms allows us to extrapolate the information extracted from choice data beyond the inevitably limited samples considered. However, this would put an enormous burden of proof on our present knowledge of underlying neural mechanisms. Is our present knowledge solid enough to sustain this burden? Although Glimcher wants to convince sceptical economists that there is already a lot that we know about neural activity, he admits we do not know much yet about what happens in earlier (‘upstream’) stages of decision-making in the brain. This is not so much of a problem, Glimcher seems to argue, because the salient stages for economists are the final ones of valuation and choice and we already know a lot about neural activity in the brain areas associated with them. Glimcher does not conceal, however, that as things presently stand, his Main Thesis that all decision-making in the brain terminates in two stages occurring in particular brain areas, is speculative and contested within neuroscience (p. 268). So the nagging doubt remains: Why would economists want to base their theories on speculative conjectures advanced in another discipline, given that a viable alternative is readily available?

These qualms do not detract from the many qualities of the book. Unsurprisingly for a book that spans several disciplines, it is a demanding read. But the book is clearly structured, most chapters start with a helpful Précis and there are many signposts that help the reader to keep track of the overall argument. Another merit of the book is that Glimcher tries to anticipate and defuse doubts about his project. This keeps the sceptical reader motivated to go on reading. The book also provides accessible overviews of recent developments in neuroscience. One of the things that neuroscience seems to have made most progress on over the last two decades is the understanding of the crucial role that the neurotransmitter dopamine plays in the learning and storing of value. How dopaminergic neurons encode a reward prediction error and are involved in reinforcement learning processes gets an accessible and thorough treatment in the book. Even though most of the things Glimcher discusses do not speak directly to the sorts of things economists and decision-theorists traditionally have been interested in, for those who have not dogmatically shut their eyes for insights and findings from psychology and neuroscience, Glimcher's book provides a fascinating and rewarding read.