Saturday, 17 November 2012

No surprises (1)

In 2010, Professor Karl Friston wrote an interesting, if somewhat technical paper about the Bayesian brain and work being done on unifying theories of brain function. In simple terms, the idea is that the human brain functions in such a way as to minimise surprises, which as Friston says are neural costs in the sense that surprises have to be dealt with by non-routine neural activity.

This non-routine activity may be a change to our mental maps of reality or a change to reality itself  - a change we make to physical environment.

We walk through a room and a chair is in our way - that is a surprise in the sense that alerts us to a mismatch between reality and route we are taking through the room - our mental map of it. So we move the chair or deny its existence - these are two of the remedies open to us.

So a surprise in this sense is a mismatch between our mental maps of the real world and our sensory inputs - which has to be unexpectedly corrected.

Not a remarkable conclusion of course, because it is consistent with survival strategies. When we adapt to an environment or situation, we act to minimise any surprises it may spring on us. When we learn and imitate others, we are acting to minimise surprises by doing what has been done before.

We do a huge amount of surprise-limitation at the interpersonal level - constantly nudging the behaviour of other people towards our own comfort zones, while at the same time being constantly nudged ourselves.

It never stops.

It can be as subtle as failing to smile or as crude as a fist in the face. It's also government policy, because all government institutions want unsurprising citizens with predictable behaviour. The corollary of course is that we ought to see this kind of government behaviour as unsurprising.  

Friston and others in his field are using this commonsense view as a mathematical model for how the brain operates efficiently, how it processes sensory inputs as efficiently as possible by keeping to a minimum those inputs which require more than routine processing – the surprises.

If surprises occur then sometimes we act on the outside world in order to minimise future surprises. We move that chair to avoid falling over it, have the boiler serviced to avoid a breakdown, drink less alcohol to avoid a headache. 

What interests me about this approach is the way Friston uses the simple word surprise to bring out how crucially important it is to minimise mismatches between experience and expectation. If his approach is sound, then this is the goal-seeking aspect of brain function at a cognitive level.

To my mind, this use of the word surprise also fits in very well with common sense – how we differ in our attitude to surprise and our reluctance to change either our ideas, our behaviour or the outside world. Although there is a loose end here, because there is such a thing as incompetence.

We attempt to minimise surprise in numerous ways. Conventions, consensus, institutions, training, teaching, instruction manuals, points of view, prejudices, braking systems, suncream, vaccinations and so on. The list is endless, because this is a very general idea.

At some point, if it hasn’t occurred already, somebody will suggest that we vary in our tendency to avoid surprise. Not that such a possibility would be surprising of course, but what if it turns out that these tendencies differ between the sexes, between races and cultures and across age ranges?

Some people may be –

  • Better at what we might loosely call administration – surprise is minimised now.
  • Better at what we might loosely call exploration – surprise is minimised in the future.

 Exploration in this sense is a search for new situations where surprise is more minimal than the present situation, from building a better mousetrap to proposing a new drugs policy to finding the Higgs boson. It's the search for better where better means more efficient, more predictable and therefore less surprising.

Obviously the potential for controversy is huge if it turns out that men and women differ, if only marginally, in their propensity to avoid surprises. If there are cultural or racial differences too? The outcry will be formidable, but why?

Because for many, the findings would count as a surprise.


Sam Vega said...

That must be why my daughter turned her nose up at her birthday presents!

It is a nice theory, in the sense that one can see all sorts of applications wherever you look.

As a lover of maps, I agree that there is a comfort and satisfaction in knowing that one's picture fits the world. But I suspect there is something higher; something that I can feel dissipating when I learn the name for that mist-shrouded copse, or realise that beautiful roof-line is the back of the old row of cottages behind the High Street.

There is a beauty in being lost, there is wonder in being surprised, and sometimes we just need to not know what is going on.

I think we get fleeting glimpses of this in Calvino, and I think this is what Blake was getting at when he railed against the systematisers and explainers like Newton.

A K Haart said...

Sam - I agree. If Friston's theory is sound, then the theory itself is attractive because it minimises surprises. It's just another pleasing map.

We never can close the loop, as Calvino knew - as I think most of us know.

Angus Dei said...

Not surprised at all AK...

Demetrius said...

I wonder if there is an age factor in that for most people the older they get the less they can cope with or deal with surprises. It would be intriquing to relate this to neural capability.

A K Haart said...

Angus - nothing surprises me. Almost!

Demetrius - probably so, yet many younger people seem to find surprises problematic too.

James Higham said...

Kick the chair over for having the temerity to be there in these self-entitled times?

A K Haart said...

James - then deny the kick!