Sunday, 11 November 2012

The Mind-Body Problem in Science

For the philosophers out there who had an aneurysm upon reading the title, just bear with me for a minute. Instead of attempting to tackle dualism using science (and thus invoking scientism to a degree that would make Sam Harris proud), I want to focus more on how naive assumptions of the interaction between mind and body can give rise to fallacious reasoning - particularly in interpretations of neuroscientific research. In other words, this is mostly going to be a rehash of articles like "Your Brain on Pseudoscience" and "The Rise of Popular Neurobollocks"; and my favourite of this genre of cranky-skeptical diatribes, an article written by Massimo Pigliucci called: "The Mismeasure of Neuroscience". 

Massimo describes the fundamental problem quite succinctly here:
Let’s begin with what exactly follows from studies showing that X has been demonstrated to have a neural correlate (where X can be moral decision making, political leanings, sexual habits, or consciousness itself). The refrain one often hears when these studies are published is that neuroscientists have “explained” X, a conclusion that is presented more like the explaining away (philosophically, the elimination) of X. You think you are making an ethical decision? Ah!, but that’s just the orbital and medial sectors of the prefrontal cortex and the superior temporal sulcus region of your brain in action. You think you are having a spiritual experience while engaging in deep prayer or meditation? Silly you, that’s just the combined action of your right medial orbitofrontal cortex, right middle temporal cortex, right inferior and superior parietal lobules, right caudate, left medial prefrontal cortex, left anterior cingulate cortex, left inferior parietal lobule, left insula, left caudate, and left brainstem (did I leave anything out?). 
I could keep going, but I think you get the point. The fact is, of course, that anything at all which we experience, whether it does or does not have causal determinants in the outside world, has to be experienced through our brains. Which means that you will find neural correlates for literally everything that human beings do or think. Because that’s what the brain is for: to do stuff and think about stuff.
What he is describing here is a phenomenon known as the 'reverse inference fallacy' and this is just a specific example of "affirming the consequent" in logic. The traditional application (or misapplication) of the reverse inference fallacy is described by Poldrack1 who presents the argument as:

  1. In previous studies, when cognitive process X was assumed to be involved, brain area Z was activated
  2. In the current study, when task A was presented, brain area Z was activated
  3. Therefore, activation of brain area Z in the current study demonstrates the involvement of cognitive process X during task A.

This can also be presented as such:

  1. If P then Q
  2. Q
  3. Therefore, P.

The fallacious nature of the reasoning can be highlighted by inserting any everyday relationship, for example: "If it is raining, then I have an umbrella. I have an umbrella. Therefore, it is raining". This is an obviously false statement as we can think of a number of situations where (accepting the initial if-then premise) I could have an umbrella without it being raining, like if my old one had broken and I had just purchased one at a store, or maybe I'm on my way to a fancy dress party where I have donned my infamous Mary Poppins costume. 

It's important to keep in mind, however, that just because the logic is fallacious, it does not mean that the conclusion is necessarily false. So even though there are possible exceptions to me having an umbrella and it being raining, it could be the case that when the argument is used, it does happen to be raining. Or, in Poldrack's example, even though it does not follow that just because cognitive process X activates brain area Z that the activation of brain area Z must implicate the involvement of cognitive process X, it could be true that cognitive process X actually is involved. I'm not sure if this coincidental correct conclusion has a fancy Latin word to describe it but I liken it to the saying that a broken clock is right twice a day; that is, the fact that the clock is broken does not justify the claim that the time it reads is definitely false, but it does justify our skepticism over the process by which it reaches the correct time twice a day. 


To be honest, it doesn't have a whole lot to do with the mind-body problem that has been speculated upon and debated by philosophers for centuries which largely revolves around the questioning of the metaphysical substances that underpin reality, but it does raise an interesting issue regarding our intuitive understanding of the mental and physical worlds. By this, I mean that even though the notion of dualism (the idea of an intangible incorporeal entity or mind that controls our actions independently of material laws and relationships) doesn't make a substantial appearance in scientific explanations and theories, it is still implicitly treated as a valid alternate explanation for behavioral phenomena through the use of the reverse inference fallacy.

This may sound unbelievable but let's look at how this argument could be applied to some recent research: "Sex, Lies and fMRI—Gender Differences in Neural Basis of Deception"2. In this study, the authors look at the problem of lying and question whether the flaws associated with previous attempts at detecting lies (like polygraphs and microexpressions) could be eliminated by cutting out the expressive component of detection and looking directly at the brain during a lie. To do this, they simply gathered together a group of people, asked them to fill out a questionnaire and then instructed them to lie some of the time when later asked if certain statements about themselves were true or not. So they would be asked questions like if their name was "John" and they would have to answer either "Yes" or "No" - with the intention of sometimes "fooling" the "lie expert" and fake equipment that was supposed to measure stress levels, when in actual fact the lie detector was the fMRI.

The actual results of the study were quite interesting, with the finding that whilst men and women used the same areas of the brain for lying, men appeared to show a greater blood-oxygen-level-dependent (BOLD) signal - possibly suggesting that there was greater effort required on the part of men to lie compared to women. The problematic aspect, however, was when the authors tried to argue that this demonstrates that women find it easier to lie due to biological, not sociological, reasons. Ignoring for the moment the validity of the results, and whether women really do find it easier to lie or whether there were any methodological issues with the study, is this a valid conclusion to reach based on the data presented?

I'd suggest that it is not, and I think it could be useful to apply the reverse inference fallacy to the basic argument behind the authors' position. The difference is that instead of fallaciously attributing the involvement of a cognitive process to a task, it seems to be a higher level mistake wherein a biological cause is assumed due to the flawed logic. In logical form, it looks like this:

  1. If a cause is biological, then it must have a neural correlate
  2. Behavior X has a neural correlate
  3. Therefore, the cause of behavior X is biological.

Like the umbrella-rain example, this conclusion also has obvious exceptions that disproves the reasoning used to reach it. These exceptions are best explained by Massimo in the linked article at the beginning but the simple point is that everything we do, think, feel, believe, and so on, is controlled by the brain. If we learn a new skill, like riding a bike or solving quadratic equations, then there will necessarily be some shift or change in the biological structure of our brains. The only way it wouldn't be true is if dualism were a viable alternative; that is, if it were true that our mental life takes place independently of physical constraints and operated entirely independently of material systems like biological organs.


Of course not (for the most part). When I talk of these scientists implicitly accepting dualism as a valid possible alternative, my intention is to suggest that this is an unfortunate side effect of poor reasoning rather than a concerted effort to consider all alternatives, no matter how silly they may be. More simply, my point is that if we (as scientists) aren't willing to accept the notion of an ethereal soul without evidence, then we have to accept that the brain does stuff. If a cognitive process, behavior, thought, belief system, etc, exists then it will inevitably have a corresponding brain area - regardless of whether the thing in question is innate/biological or learnt/environmental.

If we aren't careful in interpreting our data then we can fall victim to circular reasoning and end up continually restating our assumptions. So, for example, if we find that different brain areas are activated in the brains of democrats and republicans, it does not necessarily mean that the difference is biologically caused and it is circular reasoning to suggest that the differential activation causes the different political positions. Before scanning their brains we know that they will be different because they hold different beliefs on politics - if the difference isn't in the brain, then where else would it be?


1. Poldrack, R. (2006). Can cognitive processes be inferred from neuroimaging data? Trends in Cognitive Sciences. 10(2), 59-63.

2. Marchewka A, Jednorog K, Falkiewicz M, Szeszkowski W, Grabowska A, et al. (2012) Sex, Lies and fMRI—Gender Differences in Neural Basis of Deception. PLoS ONE, 7(8): e43076.


  1. Excellent article! I wish researchers were more cautious when interpreting the results of neural correlation studies. I don't think they have more power than behavioral experiments for the purpose of establishing causes of behavior and for testing cognitive models I think that even with neural evidence the hypotheses remain underdetermined by the empirical data...

    1. Thanks Asta. I agree, I think there is a pervasive belief that an explanation for a behavior is "more real" if it is grounded in biology. So saying that a behavior becomes more probable when it is followed by reinforcement is taken less seriously than someone saying that a behavior becomes more probable when certain synaptic connections are strengthened - despite the fact that the two explanations mean the exact same thing.

      And, as you correctly mention, when attempting to establish a cause of a behavior, looking at the neuroscience will often have less power than using a behavioral account as the neuroscience will approach the problem from the wrong level of explanation.