But we seem to get by. We do manage. And the reason that we manage is actually captured in a theorem that was developed by this 18th century Presbyterian minister Thomas Bayes. Many of you’ll have heard of Bayes’ theorem. You certainly come across it more and more in neuroscience nowadays. And Bayes pointed out, and in fact, Bayes, although he didn’t use the term, was talking about this form of abductive inference, about working out the causes based on the evidence.
And he suggested that the optimal way to do this is to take the evidence but also to take what you already know, what your prior experience and expectation is, and fit those together, and resolve the ambiguity like that.
So imagine you’re walking down a country lane – let’s say, in Devon, and you hear the clip-clop of hoof-taps up around the band. You confidently expect that what will come around the corner next is a horse. But actually the evidence you have doesn’t tell you that necessarily. I mean who can tell the hoof-taps of a horse from a zebra, or a camel, or somebody banging a couple of coconut shells together? The evidence itself is ambiguous.
But of course, it’s your prior knowledge that what’s most likely to be around the corner in Devon is a horse not a zebra. And that’s essentially the insight provided by Bayes’ theorem and the notion of the inference that we apply to the world.
There’s good evidence that this sort of inference also occurs at lower unconscious perceptual levels, and we’re doing it all the time. Looking at this image here, to many of you who haven’t seen what I’ve seen, a meaningless collection of black and white blobs. To me it’s not, to me it’s a very, very meaningful picture.
I can see a lot in it. I can see there’s a woman. I can see that she’s young. I can see she’s wearing a hat. I can see that she’s happy. And I can see that she’s kissing a horse. And the reason I can see all that is not necessarily because I am hallucinating – although there’s an element of that actually, I think – but it’s because I’ve seen this image.
And this image was the original image from which the first one was created. Now, having given you that prior information, that prior expectation, it becomes more possible that some of you may now be able to look at the image and see the woman kissing the horse. This also works in the auditory domain.
If you listen to this… (ambiguous sound) to many of you that will sound like a sort of meaningless bird-songy type of sound. To me actually, it’s highly meaningful. The reason it’s meaningful to me is I have prior expectations and prior knowledge. My prior knowledge comes from having heard this, (voice over) “The camel was kept in a cage at the zoo.” So now you have that prior experience and knowledge.
Will you be able to apply it to the original sound? (Ambiguous sound) It’s very, very striking, just how automatic and easy it is to now make sense of what was previously noise merely because of what you brought to the table. So that’s great.
We’ve got a means of dealing with the world, which allows for the ambiguity of our incoming messages, allows us to assess what the cause is likely to be, by using this inferential process based on prior knowledge. But that should give us pause, because it’s telling this: perception is an active process. It’s not a passive, a receiving of a veridical world out there.
And if the process is active enough to allow us to suppress the noise, recognize the signal, remove the ambiguity, is it also active enough to create perceptions? We are very prone to creating our own perceptions.
So here we see a famous triangle illusion, where most of us will have a very strong sense that there’s a white triangle super-imposed upon these black shapes. The reason we see that probably is because the way the shapes are arranged below it seems to strongly imply that there is a white triangle, and therefore, we create it.
There is actually nothing objectively there. There is no border here. There is no perimeter. We’re seeing something that’s objectively not there. And it’s only 10 minutes ago that I defined that as a hallucination. So this is a hallucination in action. And as Von Helmholtz described it, perception itself is controlled hallucination.
So we have this situation where we have a balance between what’s coming in and what we already know. Under some circumstances, we don’t have any strong expectations, in other circumstances, we have these prior expectations, and we will weigh them.
And this offers us a mechanism for beginning to understand the emergence of phenomena like hallucinations, because it suggests that we don’t need to hypothesize some gross derangement of function, some horrible lesion somewhere. We all define ourselves according to our internal models of the world. We define our place in relation to others, in terms of shared models of the world.
And if somebody’s building a model that isn’t shared by other people, that’s a very, very isolating experience, because their reality becomes different.
Now, I think science can come some way towards trying to look at mechanisms and develop clinical ideas of that, but ultimately, as is encompassed in the theme of today’s talks, really we need to think about experiences like this at other levels.