Robert McMaster: “neuroeconomics’ empirical basis is fundamentally underdetermined”
Robert McMaster’s paper “Neuroeconomics: A sceptical view” in the current issue of RWER outlines neuroeconomics claims and scrutinises them from a non-neoclassical perspective. He argues that “neuroeconomics’ empirical basis is fundamentally underdetermined” and that “the framing of ‘economics in ‘neuroeconomics’ is profoundly reductionist and incapable of providing the explanatory depth its proponents claim.” McMaster’s critique focuses on the work of Colin Camerer, P. W. Glimcher, D. Ross, M. Vercoe and P. J. Zak. Below are three key passages from the paper. Emphasis in bold has been added by the blog editor.
The foregoing, however, masks the evolution of neuroeconomics along potentially divergent paths, from Camerer et al’s variation of “behavioural economics in the scanner” (Harrison and Ross, 2010) a possible challenge to rational choice, to Glimcher’s (2003) variation, which Ross (2005) has termed “neurocellular economics”, supportive of rational choice. Zak’s (2004; 2008b) neuropharamlogical approach, which presents markets as sites of morality, offers a further dimension.
Given this, there may be doubts over the coherence of neuroeconomics that may undermine claims to present a unified theory of behaviour. Fumagalli (2010), for instance, outlines a “labyrinth” and “panoply” of differences (see also Harrison and Ross, 2010). By contrast, Vromen (2008) considers fracture points as exaggerated with neuroeconomics possessing a fixed reference point: rational choice. Nonetheless, potentially significant differences appear to be evident at two levels – neural architecture and departures from utility maximisation.
Ross’s (2005) “neurocellular economics” theorises brains as distributed information-processing networks.
He considers the (mammalian) brain controls behaviour through learning about associations between reward predictors and categories of actions, and hence in this way brains and markets are claimed to share an important property – they are parallel processors of information and valuations. Ross believes the brain network can be readily modelled by constrained maximisation game theoretic experiments and simulations (see Glimcher, 2003).
Arguably, Ross’s analytical entry point is the notion of multiple selves: individuals are collections of sub-individual optimising neurons that symbiotically interact in co-ordination games (Davis, 2011). Thus, a person is the result of both intra and inter personal (evolutionary) games; they are “sculpted” and “re-sculpted” over time with language playing a prominent role (Ross, 2005). Ross (2005: 248) argues that neuroscience instructs us that neurons and neural structures demonstrate the property of servosystematicity, which in essence refers to the ability to maintain themselves as relatively autonomous entities. After all, arguably the starting point of neuroscientific approaches is the segmented brain structure. Thus, for Ross agency resides in the optimising neuron.
This is also the case with Glimcher (2003), who rejects the competing systems frame of Camerer et al. In this Glimcher acknowledges the influence of the computer scientist David Marr. In modelling neurobiological architecture Marr considered that the most obvious entry point was to pose the question as to what a particular architecture was attempting to accomplish. Thus, (mainstream) economics permits the specification of the computational goal – (expected) utility maximisation – of the brain as it furnishes the benchmark for survival and reproductive success in human and non-human species. Indeed, in using experimental analysis of monkeys’ choice behaviour and the behaviour of individual parietal neurons Glimcher (2003) offers the prospect of a literal application of the rational choice model. Utility
maximisation occurs at the neural level and hence throughout the brain. Pace Camerer et al, Glimcher and his colleagues contest that there is no evidence to substantiate the case for two “independent” systems – emotional or irrational and rational – within the brain. Indeed, in his book, Decisions, Uncertainty, and the Brain, Glimcher (2003) makes no reference to emotions.
Thus, the rationality debate within neuroeconomics is partly derived from seemingly divergent ontological positions regarding the structure of the brain. For instance, Camerer et al’s dualistic systemic view of the brain’s structure adopts a localisation position that there are distinct decision-making modules within anatomical regions of the brain arising from differing evolutionary origins (see also, Zak, 2011. Contrastingly, Glimcher promotes a monistic ontology; conceiving a unitary neural structure, which is shaped by evolution in a fashion that promotes a unified behavioural pattern tailored to maximise reproductive success given environmental conditions.
Then there are Zak’s (2008b) and Ross’s (2005) invocations of the market. As noted, Zak makes the claim that markets are moral and promote morality through the enhancement of the “moral molecule”, oxytocin. This is all the more remarkable given the on-going financial crisis that commenced in 2007. Moreover, following DeMartino’s (2011a; 2011b) insightful analysis of ethics and the economics profession, Zak’s allusion seems to be founded on a “maxi-max” position, predicated on utilitarian and perhaps utopian perspectives that relegate individual autonomy and the ethical imperative of professionals exercising care, “to avoid preventable harm” (DeMartino, 2011a: 151). The maxi-max principle emphasises selection on the (consequentialist) basis of superiority of outcome relative to other possible courses of action. It does not, however, adequately account for uncertainty or deontology (DeMartino, 2011a). Given DeMartino’s argument, Zak’s references appear both naïve and crude.
Veblen’s engagement with psychology over one hundred years ago seems remarkably prescient given recent neuroscientific endeavours. Veblen’s emphasis on habit and a stratified conception of the mind provides a compelling analytical entry point in appreciating the intimate and complex relationship between habits, emotions and institutions. Again, by partly enabling, constraining and constituting the individual, institutions also influence human emotions, and vice versa (Wolozin, 2005). Yet whilst Camerer et al explicitly recognise emotions via their conceptualisation of an affective system they do not key into social referents and embeddedness and treat emotion as conceptually oppositional to cognition; there is much work suggesting otherwise.
Given the foregoing, I believe there is a case to question the veracity of neuroeconomics’ epistemic claims. This, however, is not the same as saying that there is no value in neuroscience and neuroscietific referents for the study of economics and other social sciences. My scepticism lies in the framing of the ‘economics’ in ‘neuroeconomics’; a framing that appears, perhaps unsurprisingly, to relegate the social and therefore generates lacuna as opposed to addressing them. It is not possible to characterise the human mind without appeal to language, and that language belongs to a linguistic community and accordingly embeds the individual into the social (see Davis, 2011; Dupré, 2005), but neuroeconomics appears if not ignore this then to treat it as a mere triviality. Vercoe and Zak (2010: 143) refer to Veblen’s proposition that economics should be an evolutionary science and advocate the construction of “inductive models that are problem driven, rather than imagination driven (as in deduction)”. It is a great pity Vercoe and Zak’s reference to Veblen is so partial; further reading would have revealed the centrality of the social dimension to Veblen’s work and to the comprehension of human behaviour.
The McMaster’s paper is downloadable here: Neuroeconomics: A sceptical view