Dr. Andy Clark is a Professor of Cognitive Philosophy at the University of Sussex who recently wrote a book called THE EXPERIENCE MACHINE: How Our Minds Predict and Shape Reality. It has been widely believed that our senses alone give us direct access to the world and what we see is what is there. But there are discoveries in neuroscience and psychology that turned this assumption on its head. What if rather than perceiving reality passively, your mind actively predicts it? Dr. Clark discussed these possibilities in his book.
To learn more about his background, Amit Chowdhry of Pulse 2.0 interviewed Dr. Clark.
Amit: Predictive processing is a compelling concept. What inspired you to make this concept the focal point of the book?
Dr. Clark: Predictive Processing is a relative newcomer on the scene, but it had the resources to explain all the things that had previously occupied me in my career as a cognitive scientist. It is a very neat picture of the core thing that the brain does, which is to use its own predictions (based on a model formed from past encounters with the world) to help perceive and act in the world. But more than that, it offered a unified picture that makes sense of so much that is otherwise odd and puzzling in human experience – for example, our susceptibility to chronic pain and (on the brighter side) the way that strategic interventions (the many ‘hacks’ that I discuss in the closing chapter) can actually help us push back against misfiring predictions so as to improve our own lives and experiences. It also offered the very systematic model of what brains do that accounts for something else that has occupied me a lot in the past – our tendency to in effect ‘merge’ with our best informational technologies, creating the larger ‘cyborg’ thinking systems that I call ‘extended minds’.
Amit: Do you see parallels between predictive processing and Pavlovian responses while considering the examples laid out in the preface of the book (phantom vibration phone experience) and the prerecorded bird noises?
Dr. Clark: Yes. Predictive processing neatly assimilates all the existing insights from associationist work, work on priming, work on visual illusions and hallucinations, and more. But it goes beyond simple associationism by showing that our predictions are based on structured models rather than simple associative links. The structured models are ‘generative’ (in that sense, like the ones behind ChatGPT and so on) and can construct artificial versions of possible sensory information using stored knowledge about patterns, elements, and regularities. So to fine-tune the picture, brains are prediction machines that use internal generative models to drive perception and action.
Amit: On pages 107/108 of the book, it points out that people with chronic depression immunize themselves from positive information. For those who may be dealing with depression – especially in the age of social media influence and political divisiveness: In your research, have you found ways to suppress this from happening?
Dr. Clark: I’m not a clinician so at those points in the book I simply draw on the expertise of others. The passage you mention is about how depressed brains often respond even when shown good evidence that things are not quite as bad as they seem – they respond by rejecting the new evidence, perhaps finding explanations such as ‘it only looks as if those people like me, but really they are just pretending’.
Predictive Processing has a rather specific story to tell about the mechanisms at work there – they are the ones that constantly try to balance our brain’s best predictions against sensory evidence. But if the balance is weighted too far in favor of a negative expectation then new positive evidence is simply treated as noise and ignored or re-framed. To push back against this, various techniques are used to try to increase the patient’s own confidence in the more positive evidence. They can include explaining the way misfiring (over-confident) predictions can cause a range of perceptual illusions and getting the depressed person to recognize that the very same mechanisms can cause these more ‘cognitive’ illusions too. Recognizing this is often a first step to breaking the vicious cycle in which negative expectations cause failures to interact well with others, which cements the negative expectations.
Amit: Could you tell me more about your background?
Dr. Clark: My dad was a policeman in London, and my mother stayed home to look after the family – four boys and a grandmother, all living at home. I was the first in my family to go to University (the University of Stirling, up in Scotland where my dad was from) where I studied French Literature and then Philosophy. But my first real job involved teaching in a Cognitive Science program at the University of Sussex, where I learned to program – old-school languages from back in the days when AI was the heart of the sciences of mind. Later, I directed the Philosophy-Psychology-Neuroscience program at Washington University in St Louis, getting in on brain action just as new neuroimaging techniques were bringing brain science deeper into the conversation. That was followed by a stint as Director of the Cognitive Science program at Indiana University, Bloomington, and then 14 years in Edinburgh (as Professor of Logic and Metaphysics) where I started a very successful program called Mind, Language, and Embodied Cognition. Now I’ve come full circle back to the University of Sussex in Brighton on the south coast of England.
Amit: What made you decide to become an author?
Dr. Clark: My mother loved words and writing, and I have always got the greatest pleasure from trying to express complex ideas in simple and elegant ways. Writing is a very safe space for me.
Amit: How did the idea for your new book come together?
Dr. Clark: I was deep inside a salt mine with my old friend David Chalmers and a bunch of other academics working on mind and consciousness. It was June 2018 and we were there as part of the 22nd conference of the Association for the Scientific Study of Consciousness, in Krakow, Poland. Chatting over the conference dinner deep underground (served with silverware and tablecloths on tables made entirely of salt) it struck me that there was all this fascinat¬ing work on brains as prediction machines (‘predictive processing’) that was still not fully appreciated by many cognitive scientists, despite being highly relevant to just about anything to do with mind, awareness, and action. The conversation then turned to my previous work (some of it with David Chalmers) on the ‘extended mind’ – the way minds like ours rely heavily on external tricks and tools, such as iPhones to sticky notes, to share the information processing load with the brain. My colleagues were unsure how work on predictive brains could fit together with that work on extended minds. The idea for the book emerged at that moment – to show how the two pictures converge, while at the same time introducing a wider audience to the core picture of the brain as a prediction machine
Amit: What has been your favorite memory in putting the book together?
Dr. Clark: Seeing the bright yellow cover for the US edition. It’s a real cracker. I love the UK cover too, but it’s a slower burn.
Amit: What are some of the challenges you faced in writing the book?
Dr. Clark: My editors were just too damn good. They made me write and re-write and re-write again until it really did all make sense in a way that aims to be both accurate and widely accessible. Also, the fact that the field is moving so fast. Every time I looked away to write stuff down, important new papers and results were published. Fortunately (so far at least) the newest results lend further support to the general model.
Amit: What are the main takeaways that you want readers to understand from the book?
Dr. Clark: First, the way we experience our worlds is shot through with our own history. This is because our previous en¬counters with the world are the source of the predictions and models that shape future experiences. This really matters because it shows how very deeply even what strikes us as simply seeing the world (or feeling pain, or hearing sounds) reflects our own individual life experiences and current expectations. So this is a new window into the hidden diversity of human experience.
Second, as this prediction-based system operates, it is constantly trying not just to predict the external world but to predict the states of our own bodies too. A lot of puzzling stuff about dramatic individual variations in the experienced levels of pain, about what lies behind many cases of ‘functional medical symptoms’ (those apparently lacking a standard physiological cause), and about the power of placebo and nocebo responses, all then fall neatly into place.
Third, the emerging picture is quite a hopeful one. Understanding more about the role of our own hidden predictions in building our experiences makes it clear that we actually have significant room to alter those experiences by altering the suite of hidden predictions. For example, using the kinds of verbal re-framing at the heart of what is called ‘pain re-processing theory’.
Finally, because of the way predictive brains are also ‘information-seeking organs’ (predicting what actions will best reveal useful information), they blur the boundaries between our own minds and whatever our best, easily accessible, well-trusted technologies can make available. Brains like that drive actions that make the most of whatever the world can offer. They are the perfect platform for cyborg minds.
If you are interested in buying this book, then you can find it on Amazon below:
Pulse 2.0 may receive a commission for purchases made through this article.
