The Neuroeconomics of Addiction — A Reply and Further Comment
The mathematical content of the Economics taught in universities has long been a topic of controversy. One of the lesser discussed repercussions of this is the almost complete absence of philosophy in general and epistemology in particular from economics lectures. Even though I am an apologist for the mathematically-oriented pedagogy of economics, I believe that much harm has been caused to the economic discipline itself as a result of the scant attention that has been given to philosophy in economics lecture rooms. So much so that even economics students themselves are confused about the content and scope of the subject.
Similar issues are noticeable in the aforementioned article. Several criticisms raised against the Becker-Murphy model in the article have the same flavour as those raised against the neoclassical economic thought. Here, I would like to respond to these criticisms and then try to address the divergence between the objectives of economics theory and of the so-called discipline of Neuroeconomics. I maintain that this divergence nullifies the case for any “neuroeconomic” theory to ever supplant any reasonably established economic theory. At best, these theories can inform the motivation behind economic models. Indeed, the arguments I set forth here are by no means novel. Most of what I argue comes from Gul and Pesendorfer, Rodrik, and Friedman. My aim is to reiterate their arguments in the present context so that the readers are informed about both sides of this debate.
The foremost criticism stated in the article is that the behaviour of real-life addicts and their addictions is not sufficiently explained by the model. That the models are not “detailed enough” or “realistic enough” are extremely common complaints. What all these critics fail to realise, however, is that no model ever claims to be a complete description of the phenomenon it tries to depict. Not even those of physical sciences, for that matter. The objective of any model is to isolate specific relationships which are of relevance in a particular context and study them. Economic models, therefore, are not attempts at postulating a complete characterisation of human behaviour. They aim at studying human behaviour and its relationship with economic variables. The Becker-Murphy model, therefore, never claimed to completely explain the behaviour of addicts. It attempts to study the choices made by addicts and how they are influenced by variables which are of interest to economics. It is then easy to see how neuroscientific data and brain physiology do not form a part of the latter.
Furthermore, as the economic models study the relationship between human behaviour and economic variables, only economic data can be used to refute these models, and not neuroscientific data. Economic models are silent on the brain and its physiology. Neuroscientific data, therefore, cannot be used as evidence against economic models, and doing so has no epistemological foundation whatsoever. The only form of data against which the economic models can in fact be tested against are the choice data (or more precisely, revealed preference data). Gul and Pesendorfer give a cogent analogy to put across this point:
“Suppose that we find that drug addicts generally satisfy the strong axiom of revealed preference in their demand behaviour. Can we argue that since addicts maximize some utility function, there are no separate brain functions and conclude then that the “limbic system” does not exist? This line of reasoning is, of course, absurd because brain science takes no position on whether choices satisfy the strong axiom of revealed preference or not. The argument that evidence from brain science can falsify economic theories is equally absurd. “
Similar arguments hold for the trite criticism against the rationality assumption. What non-economists do not appreciate is that the rationality assumption comes in many forms in the neoclassical thought, and pertains to something very specific. What may be generally thought of as “irrational behaviour” might not be interpreted as such in context of some economic models. It is not a convenient simplification, but a framework for studying economic behaviour. It puts certain restrictions on the agents in the models to facilitate meaningful analyses. Several types of behaviour which may otherwise be perceived as irrational can be rationalized by economic models. Indeed, a significant portion of what we call behavioural economics does this as well. The rationality assumption therefore gives context in which economic models are put forth. What would be a convenient simplification, on the other hand, is doing away with the rationality assumption. In that case, any behaviour may be explained by carefully manipulating the model. Economics would then become trivial, tautological, and wold not contain any “theory” in the real sense of the word. Moreover, even if we agree that a weakening of the rationality assumption is required, it is possible to do so (and has often been done successfully) within the standard framework, and this does not call for rejecting standard economics for neuroeconomics.
Different fields of study often have different interpretations of notions which share a common word in language. Utility is one such word. Even within the economic discipline, the notion has been a controversial one. In the contemporary mainstream economic thought, utility is determined by choice. If a utility function exists, then choices made by agents can be thought of as a result of maximisation of the said utility function. That is all that economic theory says. Notions of happiness, reward etc. that may be attached to the notion of utility, as understood in economics, are just expositional devices used in classrooms, and must not be taken literally. As far as economics is concerned, what the agent “feels” about his or her choices is irrelevant. What matters are the choices that are made by the agents! “Value” is therefore not at all the neurological equivalent of utility. These two are simply notions with different connotation across different disciplines.
As far as Ole Rogeberg’s remarks are concerned, one needs to understand that even if they are well taken, they still do not endorse neuroeconomic theories in lieu of standard economics. There is definitely a case to be made against misuse of mathematics in economics, but in context of assessing the scope and relevance of neuroeconomics, it is merely a strawman. Suppose for a moment that “economists focused in the area of Neuroeconomics (successfully) develop suitable frameworks that extensively as well as adequately capture the neural mechanics of an addicted mind”, as the author hopes while concluding her article. What we need to acknowledge is that a model which “adequately capture(s) the neural mechanics of an addicted mind” would not have anything to do with economics at all! This is because it is trying to explain choice as a function of brain physiology, variables on which the economic science remains (rightfully and modestly) silent. It would simply be a neuroscientific theory, which is being forcefully (incorrectly) given an economic interpretation.
It is important to appreciate the fact that the above neuroeconomic exercise is different from how economists use mathematics. This is how things go (or rather, should go): One starts with the primitives, (assumptions, axioms, etc.) which reduce economic notions to mathematical objects. Once we have these mathematical objects at hand, we can rely on the logic of mathematics to reach conclusions. These conclusions can then be given an economic interpretation. This gives us an economic theory expressed in mathematical language. The neuroeconomic paradigm turns this exercise on its head. It starts with neuroscientific notions, and reaches conclusions which are then given an economic interpretation. Any neuroeconomic theory therefore is simply a neuroscientific theory with an economic interpretation foisted on it.
I am not arguing that such endeavours are useless. However, the only way in which something like this can be relevant for economics is when such insights are used as motivation behind economic models, a practice which has long been a part of the modus operandi of serious economists. “Neuroeconomics” can therefore never be a substitute for standard economics, but a complement at best. Economic theories can definitely benefit from advances in measurement of brain functions to which neuro-economists allude to, but in a very different manner than the article proposes. Insofar as economic policy is concerned, there is no direct relevance of such theories. However, they can inform the economic models, in particular their primitives, which would then make the policy implications of the economic models more accurate.
I conclude with two remarks. First, the “active areas of research” in neuroeconomics to which the article refers are active areas of research within standard economics in their own right as well. “Learning”, “discounting”, and “self-control” have been getting attention from economists all along. However, how economists understand these notions and how neuroscientists do so may or may not be the same. Which connotation the “neuroeconomist” should use while talking about them is therefore not a trivial question at all.
Finally, one should always resist the temptation to accord to the economic discipline more importance than it has set forth for itself. Standard economics does not have any therapeutic ambitions and it does not care about whether the agent is happy (in the sense that neuroscience understands happiness). Not only should economists think of themselves as dentists, as Keynes once said, but non-economists who study the subject should also understand it as such.
The author of this piece is Tanay Raj Bhatt who is a graduate of the Delhi School of Economics (class of 2020) and is currently working as a Teaching Assistant at INSEAD, France, and as a Research Associate at ISI-Delhi.
