Do you understand? Part I. Nominal Trap and Substitution Bias

I am an experimental psychologist seeking to understand the mind by looking at the brain. What do I mean by that? What does it mean to understand something? How does one define explanation and understanding?

Please beware that I generally dislike arguments about definitions and I have no aim to subject you to a pseudo-philosophical discussion about epistemology. Having said that, I do think that these questions are worth pondering from time to time. After all, anyone can claim that they understand something, and there ought to be a way how to judge to what extent the claim is true or not. Did the high priest in ancient Greece have an understanding of the sea when he claimed that it obeyed the will of Poseidon? Did the lover understand infatuation when she swore that she was stricken by Cupid's arrow or that there was chemistry and sparks between her and her inamorato? How about the old Czech weatherman who explained that it would rain because his weather-forecasting treefrog had climbed a ladder earlier in the morning? Did he understand weather forecast? How about the treefrog?

I would dare to guess that we may agree that the questions are to be answered in the negative, since we would not consider the scenarios as instances of proper understanding or explanation. If that is the case, then what is proper understanding and explanation? In this series I want to discuss this question in the domain of psychology and neuroscience with a particular focus on David Marr's view of how to think about it. But before we do so, let us consider two examples of what understanding and explanation are not. I shall call them the nominal trap and the substitution bias.

The Nominal Trap

The nominal trap is an example of circular reasoning and refers to a situation when instead of explaining a phenomenon, we give it a name, and the name we subsequently take to be its explanation:

-Why does Jim look so good in photographs?

-Well, because Jim is photogenic.

-Oh, I see. Thank you for the explanation. I understand now.

This example given by Daniel Dennett in Consiousness Explained illustrates how a mere label, such 'being photogenic', may serve, at least on an intuitive level, as a perfectly valid explanation. However, it is important to realise that once we probe this explanation we find that we have not learned anything about the world, we have just restated the problem in new terminology. It is like my four-year old cousin explaining why her cheesecake tastes good: Because it is delicious. The problem hasn't been solved and the question hasn't been answered. The hole in the ceiling has been painted over, but not mended.

Similarly, we have to realise that the same applies for other labels we use in our reasoning and theories. To say that a patient has hallucinations and delusions because he is schizophrenic is on par with the non-explanations presented above. I don't think that any reasonable psychiatrist would take a mere label as an explanation for a complex mental health condition, but it is important to bear in mind that the label is there to help us reason about the condition and bring us closer to its explanation, rather than to allow us to bypass the hard task of explaining and lull ourselves into a false impression of understanding.

The Substitution Bias

The nominal trap is an instance of a broader category of fallacies called the substitution bias. The idea is very simple - here is how Daniel Kahneman defines it:

..when faced with a difficult question, we often answer an easier one instead, usually without noticing the substitution.

Kahneman uses the term in the context of decision-making. Imagine you intend to buy a new smartphone. If you are a techgeek who knows every jot and tittle of the technicalities of each model, you probably just choose the criterion that you deem most important and decide accordingly. If, on the other hand, like me you have little idea of what is out there, you might decide to swap the question "what is the best smartphone for me?" for another one, such as "which smartphone feels right?" or "which looks the best?" or you just buy the iPhone without thinking too much about it. 

There is no reason why the substitution bias should be limited to decision-making. In science, it may very well apply to our reasoning about explanations and understanding. In fact, earlier this year Krakauer and colleagues argued that it is precisely the substitution bias that plagues cognitive and behavioural neuroscience. In their paper Does Neuroscience Need Behaviour? Correcting a Reductionist Bias, they tackle the question of what should count as understanding and explanation, and conclude that a simple correlation of neural activity with behaviour is not it. And nor is localisation. In their words:

Deep and thorny questions like "what would even count as an explanation in this context," "what is a mechanism for the behavior we are trying to understand," and "what does it mean to understand the brain" get sidelined. The emphasis in neuroscience has transitioned from these larger scope questions to the development of technologies, model systems, and the approaches needed to analyze the deluge of data they produce. Technique-driven neuroscience could be considered an example of what is known as the substitution bias.

Consider the lover from the introduction. Did she really understand love when on second attempt she said that it was all about dopamine? Does the neuroscientist have an explanation of empathy now that she found out that oxytocin plays a role? As we know, hippocampus is definitely crucial for memory since if you cut it out you won't be able to remember events anymore (as patient HM found out the hard way). But how closer to an explanation of what memory does did it bring us? Based on that finding, do we now know how to improve memory or how to design a computer to have a human-like memory? No, we don't. We have merely substituted a difficult question with a simpler one.

I have no intention of belittling the significance of the aforementioned findings, nor diminishing the appeal of neuroscience. Neuroscience aims to study the most complicated natural computer in the world and perhaps even in the knowable universe, and by doing so it promises to unlock the secrets of what it is that makes us who we are. For that reason only, it deserves all the sheen and glory it is given. However, at the same time I do have a feeling that the field may have been resting on its laurels a bit more than it should have, or at least that its celebration has been premature.

To prove itself, neuroscience needs first to start by asking the question: What counts as an explanation? What does it mean to understand how the mind/brain works? In this respect, I fully agree with Krakauer and colleagues, and I shall discuss their paper next time.