The sense that we are a unified, consistent 'self' seems to be fundamental to our experience of the world. But all of the varied aspects of being oneself can be understood as forms of perception, writes Anil Seth.
I wake up, open my eyes, and a world appears. It is a familiar world – more so than usual in these days of semi-lockdown – but even more familiar is the experience of ‘self’ – of being me – that glides into consciousness at more-or-less the same time. This experience of selfhood is so mundane that its appearance goes by entirely unnoticed, unless actively paid attention to. We take our selves for granted, but we shouldn’t.
SUGGESTED READING
Cutting edge science at HowTheLightGetsIn Global
By
How things seem is not how things are. For most of us, most of the time, it seems as though the self is an enduring and unified entity, an essence, a unique identity: the recipient of wave-upon-wave of perceptions, and decision-maker-in-chief about what to do next. We sense, we think, we act. This is how things seem.
How things are is very different. The story emerging from a rich blend of philosophy, psychology, and neuroscience is that the self is not ‘that which does the perceiving’. Instead, the self is a perception too. Or rather, it is a collection of related perceptions. Experiences of the world, and of the self, are created by the brain following a common principle – a principle of ‘best guessing’, or what we might call ‘controlled hallucination’.
There are many ways we experience ‘being a self’. Loosely, these can be arranged into a kind of hierarchy. This hierarchy begins with low-level experiences associated with the physical body.
The bedrock idea is simple. Raw sensory signals are ambiguous and unlabelled. Although they reflect really-existing properties of the world, they do so only indirectly. The eyes are not transparent windows from a self out onto a world, nor are the ears, nor are any of our senses. The perceptual world we encounter – a world full of well-defined objects with various properties like shape and colour – is created by the brain through a process of inference, of under-the-hood neurally-implemented probabilistic guesswork. When I see a red coffee mug on the table in front of me, this is because ‘red coffee mug’ is the brain’s best guess of the hidden and ultimately unknowable causes of the corresponding sensory signals. When I experience the glow of a sunset, or the sharp taste of an adventurous cheese, that too is a perceptual best-guess. We never experience sensory signals ‘in-the-raw’. Every experience is an interpretation, a construction.
How does the brain reach these ‘best guesses’ that underpin what we perceive? An increasingly influential idea, these days most commonly known as ‘predictive processing’, is that the brain uses sensory signals as a kind of ‘prediction error’ by which it continually calibrates its perceptual predictions. On this view, perception involves two counterflowing streams of signals within the brain. There is a ‘top-down’ or ‘inside-out’ stream which conveys predictions about the causes of sensory inputs, and which cascades down through the many hierarchical levels that define each perceptual system in the brain (vision being only one example). At each level these predictions are met by ‘bottom-up’ or ‘outside-in’ prediction errors, which signal the difference between what the brain gets and what it expects, and which bottom-out with the sensory input itself. By updating top-down predictions to minimise these prediction errors, everywhere and all the time, the perceptual system as a whole comes to settle on an optimal (‘best’) guess of what is giving rise to its sensory inputs. This best guess is what we perceive.
Put this way, perception comes from the inside-out just as much, if not more, than from the outside-in. This is why I like to think of perception as a kind of controlled hallucination. Our brain is continually generating best guesses about what’s out there, but (under normal circumstances) these best guesses are reined in by – controlled by – sensory signals coming from the world. By the same token, what we typically think of as hallucination – perceiving something that isn’t there – can be thought of as ‘uncontrolled perception’.
Now it is time to return to the self. The key insight is that this process of perceptual inference doesn’t only apply to the world around us. It also applies to the self itself. The experience of ‘being you’ – or ‘being me’ – is a collection of perceptual best guesses, of internally directed controlled hallucinations.
It is at these basal levels that experiences of selfhood come to rest on the ground state of being alive.
There are many ways we experience ‘being a self’. Loosely, these can be arranged into a kind of hierarchy. This hierarchy begins with low-level experiences associated with the physical body, through to experiences of seeing the world from a particular first-person perspective, experiences of intending to do things and causing things to happen (what we might call experiences of ‘free will’), all the way up to experiences of being a continuous person over time within a rich social and cultural environment, an experienced self with a name, an identity, and a set of memories.
My contention is that each of these ways of being a self is best explained as a particular form of perception, and each of these forms of perception is underpinned by predictive processing. But now, instead of perceptual predictions being reined in by sensory signals coming from coffee cups and cheeses out there in the world, they are controlled by sensory signals to do with the self and the body.
Let’s drill down to those aspects of self that are most directly tied to the physical body. As part of ‘being me’ I experience both having a body, and being a body. The former simply highlights the observation that my experiences of things in the world are divided into ‘my body’ and ‘not my body’. The feeling that a particular assemblage of flesh and bone is ‘my body’ is called the ‘experience of body ownership’. We know not to take these experiences for granted, since they can go awry in all sorts of ways in psychiatric and neurological conditions. In the peculiar condition of somatoparaphrenia, a person may experience one of their own limbs as belonging to someone else (and as a result they may sometimes throw themselves out of bed).
The experience of being a body is less easy to define. I think of it as a formless, shapeless, inchoate feeling of simply being a living organism – partly expressed through emotions and moods, but at its deepest layers without any describable content at all besides what the neuroscientist Christof Koch might call ‘the feeling of life itself’. It is at these basal levels that experiences of selfhood come to rest on the ground state of being alive.
Both of these aspects of embodied selfhood can be understood as forms of perceptual best-guessing. When the brain is figuring out what-in-the-world is part of its body, it is combining prior expectations about the physical extent of its body – a body ‘schema’, if you will – with incoming sensory data, to update the corresponding perceptual predictions. Critically, the relevant sensory signals now extend beyond the classical sensory modalities such vision and touch to include other, lesser-known, channels such as proprioception – which signals the position and movement of the body in space. Experiences of body ownership, in this view, are controlled hallucinations in which body schemas are continually adjusted by sensory prediction errors coming from the physical body.(You might have heard of the ‘rubber hand illusion’ - a famous experiment that purports to show how malleable experiences of body ownership can be, given changes in sensory stimulation. However, recent research from our group, led by Peter Lush, indicates that the rubber hand illusion is also not quite what it seems.)
In my anti-Cartesian view, the essence of selfhood – wherever it is found – is always based on a fundamental physiological imperative to stay alive.
Experiences of being a body are also self-related perceptual predictions, but now another form of sensation and perception comes to the fore: interoception, the perception of the body ‘from within’. Interoceptive signals tell the brain about the internal state of the body, they convey its physiological condition: what blood pressure levels are like, how the heart is doing, and so on. These signals indicate how good a job the brain is doing at its most important task: the task of keeping the body alive. As with all forms of perception, the brain uses interoceptive signals to update perceptual best guesses. What do we perceive when the brain updates interoceptive predictions? The idea I’ve been proposing for many years now (along with other neuroscientists such as Lisa Barrett-Feldman) is that we perceive emotions. Emotions are also forms of perceptual inference: specifically, of interoceptive inference.
This perspective can explain why an emotional experience is different from, let’s say, a visual experience of a coffee cup. Visual experience of objects (like coffee cups) are concerned with figuring out what’s there, so it makes sense for the corresponding perceptual best guesses to have the character of ‘things’ with specific locations and physical extents. But emotional perceptions are not concerned with where things are or what shape they are. Interoception is not about figuring out where the liver is, or what colour it is. Interoception cares about how how good a job the liver is doing at filtering the blood. Emotional experiences, and experience of ‘being a body’ more generally, therefore do not have the character of objecthood – they instead have ‘valence’. Things are good or bad, or likely to be good or bad in the future.
The upshot of all this is that the deepest layers of selfhood – these basal experiences of ‘being a body’ – are intimately tied to our flesh-and-blood nature as living creatures. This conclusion upends some very old views in philosophy, often traced to Descartes, which have tried to drive a wedge between life and mind. In asserting that non-human animals lacked the essential elements of human selfhood, he called them ‘beast machines’ – flesh-and-blood automata without moral status. In my anti-Cartesian view, the essence of selfhood – wherever it is found – is always based on a fundamental physiological imperative to stay alive. We experience ‘being a self’ because of, not in spite of, our beast machine nature.
With these building blocks in place, a much longer story can be told about how the remaining aspects of self – and indeed our wider perceptions of the world – inherit from the origins of predictive perception in physiological regulation. Even experiences of ‘free will’, which are often the final refuge of those who resist the naturalisation of mind and consciousness, can be readily explained as perceptual best guesses of a particular kind. (I won’t dwell on this rich topic here. I will only hint at a parallel with experiences of colour. Both experiences seem to pick out a really-existing feature of objective reality: a property of objects in the case of colour, and a property of mind in the case of ‘free will’. But neither do. Rather, both forms of experience represent really-existing things or processes in ways that are useful for the organism.)
For me, understanding selfhood this way has been transformational. It drives at the very heart of how we see ourselves painted into the canvas of nature. Everything that it means to be me emerges from a fundamental process by which the brain generates and deploys predictive models in order to maximise the likelihood of staying alive. None of this undermines the value or ethical status of selfhood, it just puts into a richer context, a context in which we are more part of, and less apart from, the rest of nature.
Anil Seth’s new book ‘Being You: Consciousness and the Beast Machine’ will be published by Faber & Faber in 2021. For updates, join his mailing list at www.anilseth.com and follow him on Twitter at @anilkseth.
Join the conversation