Do you understand? Part II. Neurophysiologist's Nirvana

If we look far into the future of our science, what will it mean to say we ‘understand’ the mechanism of behaviour? The obvious answer is what may be called the neurophysiologist’s nirvana: the complete wiring diagram of the nervous system of a species, every synapse labelled as excitatory or inhibitory; presumably, also a graph, for each axon, of nerve impulses as a function of time during the course of each behaviour pattern ... [However] real understanding will only come from distillation of general principles at a higher level, to parallel for example the great principles of genetics — particulate inheritance, continuity of germ-line and non-inheritance of acquired characteristics, dominance, linkage, mutation, and so on. It seems possible that at higher levels some important principles may be anticipated from behavioural evidence alone. The major principles of genetics were all inferred from external evidence long before the internal molecular structure of the gene was even seriously thought about.

—Richard Dawkins

 

One thing that you often hear neuroscientists lament about is the lack of resolution of their techniques. The idea behind the lamentation is that if we could only capture brain activity just as it is happening in space and in time with the level of detail down to the scale of individual neurons, or even smaller to dendrites and ion channels, we would be but a tiny step from complete understanding. In other words, what stands between us and explanation is the coarseness of our technology and the lack of access to the details of the biological tissue. Perhaps for that reason, novel techniques that promise greater access, such as optogenetics or connectomics, are revered and celebrated so much. However, before we join the wave of enthusiasm, we ought to raise the following question: is this the right way towards understanding the mind and the brain?

While it may be the case that technology has driven the field in recent years further than any other factor, it may also be true that it resulted in a shift of attention from theory to methodology that has undermined rather than facilitated the progress of neuroscience as well as the related cognitive and behavioural sciences. While it is true that without technology we may have never been able to answer the questions that we now can, it is equally true that without good theories we do not have interesting questions to begin with. Scientific progress, we have to bear in mind, depends on the former (the state of our technology) as well as the latter (what kinds of questions we ask). In the words of Carl Woese:

Without the proper technological advances the road ahead is blocked. Without a guiding vision there is no road ahead.

A prime illustration of this virtual truism is the cognitive revolution of the late 1950s. The invention of the computer played a  considerable role in inspiring the intellectual movement, but it was the ideas, not the techniques, that turned the world of psychology on its head. Back then it was accepted that there was nothing more practical than a good theory, and I don't see a reason why we should believe that today is any different.

It is this notion that Krakauer and colleagues revisit in a paper published earlier this year called Neuroscience Needs Behaviour: Correcting a Reductionist Bias. The authors argue that currently we are pursuing a research program that primarily focuses on manipulations at the level of brain tissue, while neglecting a careful investigation and theoretical decomposition of the behaviour at hand. Since research has costs that need to be paid in the currency of money, time, and graduate students, it is worthwhile pondering whether this approach is the most effective in getting us closer towards understanding.

Imagine that we have succeeded in finding the first full human connectome and could measure the activity of each of the 86 billion odd neurons down to the nearest millisecond, would we understand what the brain does and how it produces the mind and behaviour? A hint of what the answer may be comes from research on Caenorhabditis elegans, the famous 1-mm transparent nematode. We know C. elegans fairly well. We know its full genome, all of the cell types, as well as the full connectome. However, as Anthony Movshon said, "[the connectome of C. elegans has] been wholly known for about twenty years but we don't have a comprehensive model of how the worm's nervous system actually produces the behaviours." The knowledge of the underlying structure seems insufficient if our ultimate goal is to understand the behaviour which the organism produces.

In order to understand behaviour we need to pose testable theories at the level of behaviour; theories that have a satisfactory level of conceptual resolution that can subsequently guide our investigation of mechanisms underlying the behaviours as well as their physical implementation. In the following part, I describe the approach and philosophy of David Marr whose aim was to tackle the problem of neural reductionism that strongly reverberates in the paper by Krakauer and colleagues. In addition to his critique of reductionism, Marr recommends an alternative route to exploring the mind and the brain that is divided into separate but adjoining levels of analysis.