What do we call this?

It is a great pleasure to write a few posts here on Brains. In this first post I will talk a little about the kind of work I like to do. Then the plan for post 2 is to ask if prediction error minimization is all there is to the mind, post 3 will look at the implication of prediction error minimization for embodied cognition, and in post 4 I plan to look a bit at consciousness from the point of view of, you guessed it, prediction error minimization.

Philosophy of mind can be many things. Nowadays it would not be totally unreasonable to call much philosophy of mind ‘speculative neuroscience’ or perhaps ‘conceptual cognitive neuroscience’.

This blurring of the lines is a good thing. Philosophers can contribute in some measure to re-formulating and re-directing traditional issues in neuroscience. It also frees philosophers from engaging in too much fine-grained metaphysics, which tends to have diminishing returns.

I am so steeped in the neuroscience that my son Lewey likes to proclaim “my dad is a pretend neuroscientist”. Not an entirely unfair label.

Over the last few years, I have built up an empirical lab here in the philosophy department at Monash. My days are taken up with lab meetings, ethics applications, and doing experiments. Our little group of PhD students, post docs and research assistants come from philosophy, neuroscience, psychology and IT. We also work with other labs at Monash and elsewhere.

I like to think that bringing experiments into the philosophy department is a good thing for the discipline. But there are some personal aspects that also matter.

After doing solitary philosophy for quite a few years, I really relish now working in a team on big, collaborative projects. It is a very different kind of intellectual endeavor.

After doing a priori philosophy for years, it was thrilling to actually discover something new in nature. (In our first rubber hand illusion study, Bryan Paton and I found that just by tweaking the sensory input a little, people can be given quite wild supernatural-like experiences.)

Doing this kind of ‘philosophy-cum-neuroscience’ is very satisfying, intellectually, socially and professionally. It can also be heart breaking, when analysis spits out p = 0.051, or when another lab beats you to a publication.

I would welcome suggestions for labels for what we do. When looking for a name for the lab we considered: “experimental philosophy”, “neurophilosophy”, “empirical philosophy”, etc. (and combinations of these). But none seem quite right. We ended with the anodyne “Cognition & Philosophy Lab”.

In the lab at the moment, we are studying how multisensory perception affects body image and action, and we are interested in how this manifests in individual differences along the autism spectrum. We are running studies of statistical learning. We look at the neural mechanisms for conscious perception. We consider how self-awareness interacts with agency. We have worked on how different considerations of risk impact how we balance fairness and equality.

These experiments look quite scattered thematically but in reality it is all driven by interest in a new, general theory of brain function, which is emerging in neuroscience (developed mainly by Karl Friston). It is the theory that the brain is engaged in prediction error minimization.

In terms of our experiments, this theory treats multisensory perception as Bayes optimal inference, action as a type of inference, autism as suboptimal prediction error minimization, statistical learning as updating priors for long term prediction error minimization, conscious perception in terms of top-down modeling of the world, self-awareness as shaped in ongoing inference, the self as an inferred cause of sensory input, and choice as a matter of evidence accumulation.

Accordingly, in parallel with all the experimental work in the lab, I work extensively on this exciting theory. This work began back in 2004 or so and led to The Predictive Mind, which came out in late 2013.

I am excited about this theory. It is a really good tool for driving and interpreting experimental work, and it has very wide-ranging consequences for many questions in philosophy.

As I see things, this neuroscience theory is centrally addressing a core philosophical question—a question of Kantian magnitude: how is it that, from within the confines of the skull, behind the sensory organs, we are able to perceive, learn about, and act meaningfully in the world when all we have to go by is the manifold of causal effects from the rest of the world (including our own bodies) on our senses?

If this question is what neuroscience is fundamentally about, then it is no wonder philosophers gravitate towards neuroscience. It means that really neuroscience just is experimental philosophy (or whatever we call this thing).

Many attacked not only Kant’s answers but also his very core question. Similarly, these days, there are serious objections to assuming this kind of sensory veil between the mind and the rest of the world. I think Kant’s question was exactly right. This is the only philosophical question that has ever had any real grip on me.

So this question is the starting point. Even if we all work on quite specialized topics in interdisciplinary philosophy of mind, our research must at some stage of inquiry be constrained by the Kantian-style question about how we make sense of the sensory manifold.

For this reason, a theory that attempts a fundamental answer to the question would be tremendously valuable.

By the way, the very short answer to the question is this: we do it by minimizing prediction error. I’ll say more about how this is supposed to work in the next post.