鈥淭HERE is no inner world. Our flow of momentary conscious experience is not the sparkling surface of a vast sea of thought 鈥� it is all there is.鈥� And with this dramatic claim, Nick Chater sets out to convince us that the ubiquitous feeling our minds have depth 鈥� that our actions and behaviours can be explained in terms of something more within 鈥� is misplaced and wrong.
In The Mind Is Flat, Chater begins by asking us to think about the suicide of the hero in Tolstoy鈥檚 Anna Karenina. We could make sense of Karenina鈥檚 actions based on what we know of her character, but Chater says that we would be making up a story because she is fictional and so has no inner life.
Real humans, he argues, also have no inner lives, and any justifications for our actions are similarly concocted. 鈥淭he very idea that our minds contain 鈥榟idden depths鈥� is utterly wrong,鈥� says Chater, a professor of behavioural science at Warwick Business School in the UK, adding: 鈥淭here are no pre-formed beliefs, desires, preferences, attitudes, even memories, hidden in the deep recesses of the mind.鈥�
Advertisement
It is a controversial argument, going against our intuition that there is more to our minds than meets conscious awareness. It also goes against current thinking in psychology and neuroscience. Chater knows this, admitting that he is disquieted by his own ideas.
Whether you agree with him or not, Chater writes passionately and evocatively. His case is that our sense of having desires, motivations and fears that drive our actions is due to the fact that thoughts are made up on the fly: there is no place in the mind where they are stored . 鈥淭houghts, like fiction, come into existence in the instant they are invented, and not a moment before,鈥� he writes. 鈥淭he stories we tell to justify and explain our own and others鈥� behaviour aren鈥檛 just wrong in detail 鈥� they are a thoroughgoing fabrication from start to finish.鈥�
He draws on psychology, neuroscience and AI to bolster his case. His favourite examples draw on visual perception, showing that our idea of a complex, colourful visual field in front of us is an illusion, the careful fabrication of an artful brain. For Chater, the sense of mental depth is also neural chicanery.
鈥淭houghts, like fiction, come into existence in the instant they are invented, and not a moment before鈥�
He also looks to the 1980s, when researchers thought they had cracked AI when they built expert systems, software specific to a domain such as health. They encoded human knowhow into a 鈥渒nowledge base鈥�, while another piece of software (an 鈥渋nference engine鈥�) used the knowledge to reason about problems.
Expert revolution
The pioneers of AI seemed to be duplicating the workings of the human mind. 鈥淭hey took it for granted that the thoughts that we consciously experience and can put into words are drawn from a vast sea, or web, or database of similar, pre-formed thoughts, which we are not currently consciously experiencing,鈥� explains Chater. 鈥淏ehind each expressed thought lies, supposedly, a thousand others beneath the surface.鈥�
Expert systems were going to revolutionise AI. They never did. And there鈥檚 a lesson in this: for Chater it means that our understanding of human minds is also wrong. We do not possess some hidden knowledge base we tap into for reasoning, to make judgements and to act; consequently it is wrong to think we can peer inside our minds to explain ourselves. 鈥淚n reality, when we decide what to say, what to choose, or how to act, we are鈥� literally, making up our minds, one thought at a time.鈥�
But Chater is setting up a straw man. Not all of us feel we are simply accessing or reasoning using a database of preformed answers to questions. And neither does computational neuroscience argue for such a database: in fact, we don鈥檛 really know the algorithms the brain uses to reason. And that is no argument for or against depth.
Chater acknowledges that the human mind is unprecedented in both its complexity and its ability. To explain its powers, he suggests that at any moment the brain鈥檚 networks of neurons are engaged in a 鈥渉ugely complex cooperative computation鈥�, a process of which we only become aware through the networks鈥� output 鈥� for example, a thought 鈥� but without ever knowing the inner workings.
In itself, this claim is not controversial: conscious awareness is not considered to be everything that the brain does, and we may never be privy to its deeper goings-on. But Chater will not allow himself to imagine that thoughts 鈥渃an be divided in two as the waterline splits an iceberg: the visible conscious tip and the submerged bulk of the unconscious, vast, hidden and dangerous鈥�. There is no iceberg.
To me, all this seems to hinge on semantics. Take what happens when he writes that there 鈥渋s just one type of thought, and each鈥� has two aspects: a conscious read-out, and unconscious processes generating the read-out鈥�. Surely if there are unconscious processes, even those we can鈥檛 access, that is evidence of mental depth? Or at least, something more than an on-the-fly model?
What do these processes depend upon? How do they give rise to our brain鈥檚 capabilities? Chater argues that our brain improvises moment by moment, and that these improvisations build on the 鈥渇ragments of past improvisations鈥�. He accepts that each of us is 鈥渁 rich store of distinctive past experience鈥�, and that there are 鈥渓ayers of precedents 鈥� the successive adaptation and transformation of previous thoughts and actions to create new thoughts and actions鈥�.
All of this sounds a lot like 鈥渢he brain has memory and learns鈥�, and that this memory and learning influences subsequent behaviour. Some computational neuroscientists would call that hierarchical deep-learning, even if we don鈥檛 fully understand how the brain pulls it off.
Humans are smart in part because we think imaginatively and in metaphors, says Chater, referring to cognitive linguist George Lakoff鈥檚 seminal book, Metaphors We Live By. But Lakoff and others have also argued that metaphorical thought is rooted in our bodies 鈥� a form of embodied cognition that is nothing if not a deep kind of mind. Chater, however, gives short shrift to this idea and to all deep-mind aspects of our being.
His flat-mind hypothesis reads like a makeover of behaviourism, the early-to-mid 20th-century philosophy that privileged outward behaviour over inner subjective states and, in its extreme form, even denied the existence of any inner mental and physiological states. Despite profound knocks from many research fields over the years, behaviourism still attracts some philosophers and psychologists.
鈥淥ur metaphorical thought is rooted in our bodies 鈥� a form of cognition that is nothing if not deep鈥�
One of the biggest challenges to Chater may yet come from AI, the very thing he thinks is unlikely to come close to replicating the human mind. 鈥淚f imagination and metaphor is the secret of our intelligence,鈥� he says, 鈥渢hen that secret may, perhaps, be safely locked away in the human brain for centuries and perhaps forever鈥� 鈥� certainly beyond the reach of AI.
The inventors of AlphaGo Zero at Google鈥檚 AI outfit DeepMind may beg to differ. In 2017, in just three days, AlphaGo Zero taught itself the game of Go well enough to defeat the previous AI champion, AlphaGo, by a score of 100 to 0. There is nothing 鈥渇lat鈥� about AlphaGo Zero鈥檚 machine-learning architecture. It is called deep learning for a reason. It is not human-style general intelligence, but rumblings are afoot that it is only a matter of time.
Book Details
The Mind Is Flat: The illusion of mental depth and the improvised mind by Nick Chater, Allen Lane
This article appeared in print under the headline 鈥淵our mind is not an iceberg鈥�
