Listening with...
Listening with...
Chris Frith on perception, swarm consciousness, and seeing things that aren’t there.
0:00
-17:27

Chris Frith on perception, swarm consciousness, and seeing things that aren’t there.

"I have a prior expectation or hypothesis. This is the evidence."

I have a prior expectation or hypothesis. This is the evidence. Should I change my prior expectation given this evidence or not?

Hello, you're listening with the Bluecoat in Liverpool, a series of podcasts taking the themes of our exhibition programmes as a starting point for 15 minute insights from artists, scientists, writers, educators, storytellers, and more.

In this episode, neuroscientist and philosopher Chris Frith discusses his work into perception, swarm consciousness, and seeing things that aren't there.

I'm Chris Frith. I'm currently an Emeritus Professor at University College and also a research associate at the Institute of Philosophy. I originally trained as a clinical psychologist, and I then went into research as an experimental psychologist, which was at a very good time, I think ’65 -  we were the first psychology department in the country to get a computer. So I started using computers very early and carried on ever since.

I then went to work for the medical research council on a special unit that was interested in schizophrenia. We did a lot of work on schizophrenia and we were particularly interested in the biological basis, where the brain comes in, but I became fascinated with hallucinations in particular and how to understand what on earth is happening when you see things that are not actually there.

I was also able to start doing brain imaging in the first unit in the country where you could do that, which was originally in Positron Emission Tomography, and then subsequently MRI: Magnetic Resonance Imaging. So I did some of the early work using brain scanners, and one of the first things we did of course is look what happens in the brain when you're hallucinating.


As you probably know, the hallucinations assisted with schizophrenia are basically hearing voices. And some people say they’re just making it up. They don't really have voices… but the brain imaging showed that there was really something going on, interestingly in the speech areas. It was as if they were speaking to themselves, but didn't realise this.

When I retired, which was a long time ago, I became particularly interested in social cognition, how people interact with one another, Schizophrenia, and my wife's speciality, Autism, seem to be in a sense, primarily a problem of interacting with other people.


I met Suki Chan through my colleague, Colin Blakemore, a neurophysiologist particularly interested in perception, and we were both working at the Institute of Philosophy when we retired, because that's what you do when you retire, you become a philosopher. Suki was particularly interested because of my work on hallucinations and perception more generally, and what we'd like to call the Bayesian approach, which is to say, actually we're all hallucinating all the time, but in most cases, what we hallucinate corresponds very closely with reality and what goes wrong in schizophrenia, for example, is that what they are perceiving no longer coincides as reality.

I'm particularly interested in artists because art - particularly visual art - is all about how perception works. My colleagues Semir Zeki has gone into this in a big way, talking about what he calls neuroaesthetics, but making the point that many of the developments in modern art, like Op art and so on followed by developments in science where you discover there’s a bit of the brain that’s only interested in movement. There's a bit of the brain that's only interested in colour. So many of these developments take account of how the brain works, which the neuroscientists only discover a few decades later. So that's quite interesting.

There are cultural effects. So we all think we see the world in the same way although we don't. One of the things we’ve noticed about synaesthesia is if you go and give a lecture about synaesthesia, somebody will come up afterwards and say, I am like that. But I thought everybody was like that. We think we're experiencing the world much more similarly than we are. What culture does is pushes us all together into this common view. And so it was very striking when the dress went viral, people suddenly found that they were looking at the same picture and seeing something different. So I think that's fascinating, and I think artists’ reveal to us how different the world actually is for other people.

The disorder in the sense that pushed me in this direction: people who lose a limb, they have an arm amputated or a leg amputated for some reason, often develop a Phantom limb. The sad thing is that in some cases they feel pain in their Phantom, and the question is how on earth does that happen?

So in a sense, when somebody has a Phantom limb, they're hallucinating where the limb ought to be if they had it, but for the rest of us were hallucinating where the limb ought to be, and it is there. So that’s the hallucination coinciding with, and also being constrained by reality. It obviously goes wrong in the case of the Phantom limb, because you haven't got one, but also it can go wrong when the constraint doesn't happen.

There's a very strange phenomenon called Capgras syndrome, named after  the neurologist who first described it. So this is where people with brain damage become convinced that their spouse is not their spouse. And they say, it looks just like my wife, but I know it isn'tjust not right. When you see your wife or anyone familiar, you have a strong emotional response alongside the perceptual recognition. And if you get damage to your amygdala, the emotional response doesn't happen so that you get this very strange experience.This is not right. My experience of seeing my wife is not quite right. And then they presume this rather bizarre account of how and why this might be that she's been replaced by an impostor. So that's where your interpretation is no longer being properly constrained. So in dementia, you'll get this phenomenon where they suddenly say, this is not my room, and the carer typically has to go round the block and then come back again and then they lose this strange experience.


There's lots of experiments where you have one person in the little room, and you present them with horribly difficult visual stimuli where they have to detect targets. It's called psychophysics and they have hundreds of trials and you see how good they are at detecting targets? So we had a twist on this, where we put two people inside this little room and they had the same task. So they had to detect difficult targets in a visual display. The same thing was presented to them separately. They made their decision and if they disagreed, they had to have a discussion and come up with a joint answer. And what we were able to show, to our delight, is that two people working together can get a better answer than the best one working on their own. So you get this advantage from working together.

It's inevitable that your attention lapses, so on some presentations, you didn't quite see it, and if you know this, then when you discuss it with your partner, you can say, well, I wouldn't see it very well on that trial, so we'll take your answer. So you can actually get rid of these bad answers, and it critically depends on talking to each other about how confident you were in your response. And that enables you to get this upgrade.


Now, what I did not realise is that bees do the same thing. So when bees swarm, they all sit in this big mass of bees and they have to find a new nesting site. A few hundred scouts go out and they explore the environment, and if they find a nesting site, they come back. So when they come back, they do their little dance, which tells you the direction of the nesting site and the length of the dance indicates how good the nesting site is.

And they assess the nesting site for… is it dry? Is it high enough about the ground? Is it big enough? There are various options and they're all doing these little dances. The idea is that they do the dance and then other scouts will go out and do look at the same site. Now, if it's a longer dance, more scouts will go out in that direction and look at that site. And if they come back and they agree, as it were in quotes, they will also do a longer dance. So eventually most of the scout bees are going to the same place and coming back and saying, this is a good site. So that's a positive feedback loop, and once they get a high enough proportion saying this is the best site, they actually then turn on the scouts who are giving different information and suppress them. And then after an hour or two, the whole swarm goes off to this new site. And that's how the bees are making this joint decision as a swarm. In a sense they're sort of using confidence.

Now, what is particularly interesting is the people who studied this phenomenon and wrote it up, the computations are exactly the same as what neurons do in the mammalian brain. So, there’s lots and lots of neurons which are getting information. They have to make a decision and that's how they finish up with a single decision. There are two things there. First of all, it suggests that the swarm certainly knows things and is able to make decisions that the individual bee can't do. So maybe the swarm is more capable than the individual bee. In that sense, maybe it's more conscious than the individual bee. But the other thing I'm fascinated by is, by putting all the bees together, you get an advance over the individual bee, and we've shown in a very simple minded way that putting people together gets an advance over the individual person.

So people working together in big groups should be able to do much better than individuals, and in fact, we know that's the case. That's where culture comes from. That's where science comes from, to some extent that's where art comes from, because it's all building on what's happened in the past. But that's why I became so interested in the swarm.

These results are too recent for them to have filtered back into the real world, but I know Professor Seeley, who has done all this work on the bees tries to organise his lab on the basis of Honeybee Democracy as he calls it.


Certainly, joint decision-making is beginning to feed into the real world because, I should point out that we also found there are situations where making decisions in groups can get worse. This is the problem that arises if some people are less competent at the task than others. So we actually did an experiment we had them in pairs again, but we deliberately made one of them less competent by putting noise into their signal that wasn't in the other person's signal. When that happens, it can actually pull it down because too much attention is given to the incompetent person.

We've done it in different cultures. We did a wonderful experiment where we did this in Iran, China and Denmark, and the same thing happens. They all equally paid too much attention to the incompetent person. And I think this is because there are different aims when you're working together. One of course, the one we’re interested in, can you get the right answer? But another one is, are we going to have a good interaction? In this case, if you just completely ignore the other person all the time, then you're not going to have a good interaction. So it’s about being a little bit fair.

So there are these interesting examples of where it can go wrong. You tend to mistrust the first person, and you want to think again. There's even a wonderful experiment where they fooled people into them giving a first answer, then they were fooled into thinking that was somebody else's first answer, and then they corrected themselves. So that would be another interesting case where you could apply that in the real world.

We make bad decisions often because the first answer that comes into our mind feels right, and there’s a great advantage of never accepting the first answer that comes into your head and thinking about it a bit.


This is the Bayesian idea that what you perceive is not what comes in - or it's partly what comes in, but it's partly what you expect. There’s always a balance between these two, and if your expectation is too great, then you'll have a hallucination.

All Bayesian theory says is perception depends on your expectation as well as the evidence. Evidence in this case being the sensory input, and one of the nice examples is the Inverse Mask effect. Now, this is an example where the whole of your life you've seen faces the right way round. you have this such a strong expectation that faces the right way round that you cannot make use of the rather minimal evidence that this thing is actually the wrong way around. And what Bayes theory does is say, I have a prior expectation or a hypothesis. This is the evidence. Should I change my prior expectation, given this evidence or not?  So the evidence is very strong. Our change is the evidence is very weak I wouldn't change it. If the prior belief is very strong, I won't change it unless the property is very weak [then] I might change it. Bayes’ equation simply gives you the mathematics of how you should make this adjustment.

Now, Bayes is a very interesting chap because he was a non-conformist clergyman who obviously couldn't do maths in England and had to do it in Edinburgh. And he became a fellow of the Royal Society, which seems reasonable given his equation is so important, but in fact, his equation wasn't published. It was published posthumously by a friend of his. So it's not clear how he became a fellow of the Royal Society. One suggestion is the reason he developed this is because David Hume had argued against miracles saying there is no evidence that would convince me that miracles can actually happen. So in fact, basis equation is trying to quantify how much evidence you would need to change a belief of this sort. That's my attempt to explain Bayes.


Memory is just like perception. So it's Bayesian. We reconstruct our memories from a little bit of information that we've retained and all sorts of more general knowledge that must been fit together. So in that sense, our memories are always incomplete and we're very easily misled, which is what barristers can do with their witnesses. The fact that we have a bad memory is what enables us to structure the world, because if we could remember everything we wouldn't need to. But the interesting thing about outsourcing memory to Google or whatever it may be - that's fascinating because we've been doing this for a very long time.

So when books were became widely available, even in Roman times people complained, if you'd write it all down, you're going to ruin your memory. It was already a big problem. I think Cicero was concerned because when you were giving speeches in court, you had to memorise them all, and the idea you would write it all down, that seemed absolutely terrible because we would destroy your ability to remember things.

So these tools are a wonderful thing and that presumably means we can release our brains to do other things, but I'm not quite sure what these other things are…

You've been listening with the Bluecoat. Produced by George Maund, Marie-Anne McQuay and Sam Mercer with sounds by Nil00. Thank you to Garfield Weston Foundation for supporting this series, and our core funders, Arts Council England, Liverpool City Council, and Culture Liverpool. Our public programmes rely on grants and donations, and you can support us at thebluecoat.org.uk/donate.

Further reading and watching

Chris Frith
Chris and Uta Frith: Two Heads
Colin Blakemore: on perception
Bayesian approach
Semir Zeki: Neuroaesthetics
The Dress
Phantom limbs
Capgras Syndrome
Professor Seeley: Honeybee Democracy