Feeling Comes First
by Aaron Penczu
Profile Books, 432pp, ISBN 9781788162838, £20.00
Neuroscience so confidently identifies consciousness with the cerebral cortex — the densely-folded, outer layer of grey matter in mammal brains — that surgeons sometimes question whether children born without one require anaesthetics at all. Yet these children cry, laugh, play, and distinguish familiar from unfamiliar stimuli — behaviours difficult to imagine alongside the total absence of experience. Animals whose cortex has been surgically removed (in grimly characteristic experiments) continue to navigate mazes, eat, procreate, and nurse their young; if anything they are more active, and more emotional, than their normal peers. Whereas lesions the size of a match head on key structures lower in the human brain obliterate consciousness, patients are often awake and talking throughout surgery which involves cutting into their cortex.
To challenge what he terms the cortical fallacy, Mark Solms also transcribes conversations with patients with supposedly key cortical structures damaged or removed, showing that — despite often very profound cognitive impairments — their sense of self and basic awareness are intact. His argument is that mainstream neuroscience has got the source of consciousness wrong. Cortical thinking is more of a plug-in than the enabling condition of awareness. Feeling comes first, and it’s generated not in the cortex but the brainstem, the most evolutionarily ancient part of our brains.
The brainstem is conventionally thought to be a kind of power cable for the mind, necessary but not sufficient, and responsible mainly for modulating arousal, as in the transition between wakefulness and sleep. Yet the neurotransmitters which most anti-depressants act on are produced in the brainstem, and ‘all the brain’s affective circuity’ converges on a tightly packed knot of neurons in it called the periaqueductal grey. This is the hidden spring of the title and, in Solms’ view, the fount of consciousness in the brain.
As well as a practising neuropsychologist, Mark Solms is an affective neuroscientist, and his book is the best of a recent slew which introduce this rival tradition, focused on emotion rather than cognition, to a general audience. By packaging together perspectives from the study of affect with more widely-known ideas in neuroscience, like the Bayesian brain hypothesis, as well as Solms’ own polemic claims, The Hidden Spring aims also to sketch an ambitious, overarching account of consciousness. It culminates with an alternative vision for how we might create a consciousness artificially – intending, unlike AI research focused on computation, to produce a subject that feels.
Solms begins by asking why consciousness exists at all. Most brain functions, like heart rate control or the conversion of electrical signals into auditory representations, occur without self-awareness because they involve stereotyped, inflexible responses. Consciousness arises, he suggests, from the need to modulate behaviour flexibly, in real time, in relation to a complicated environment and competing biological needs. The part of the brain controlling volition is guided in this task by computations of biological value, eloquently expressed as affects.
Imagine being inside a pub cellar flooded with carbon dioxide from a leaking cylinder. The gas is invisible and odourless: it’s the continuously varying feeling of suffocation that might guide you to breathe oxygen-rich air higher in the room. The sensation’s aversive ‘hedonic tone’ drives you to avoid or eliminate it, whereas positively-valanced experiences, like eating or sex, tempt us to extend them. The brain can only survive undamaged without oxygen for a few minutes, so the feeling of asphyxiation is much more intense than hunger or thirst.
Hedonic tone colours almost all of our experiences, and undergirds a simple but ubiquitous form of associative learning. Even a patient unable to form new memories avoids shaking hands with her doctor the day after he conceals a drawing pin in his, though she cannot recall ever meeting him. Like her, Solms suggests, we may not always know why we feel something, but are always aware of feelings themselves. Their function is to express the nature and urgency of biological needs — protection from injury and infection, in the case of the amnesiac patient — and their mechanism of action is consciousness, that is, they work by being felt.
The science of emotion is in some respects still in its infancy: there’s no generally-accepted taxonomy for affects and feelings, and foundational aspects of their neurobiology are disputed. Some researchers believe emotions are socially learned constructions, and the only constant across humans is the hedonic core of aversive and attractiveness. Solms sides with those who see emotions as biologically intrinsic not only to humans but also other animals. Following the late Estonian neuroscientist Jaak Panksepp, he distinguishes interoceptive affects such as nausea, fatigue, or the need to urinate from outward-facing, exteroceptive affects like pain, disgust, and surprise, and the less body-centric emotional affects, including jealousy and love.
Panksepp, perhaps the central figure in affective neuroscience, studied emotions in a ‘veritable menagerie’ of animals, including pigeons, chickens, beagles, guinea pigs, rats, and prairie voles, which he sacrificed to science in droves. His methods included lesion studies, pharmacological challenges, and deep brain stimulation, a process involving delivering electrical impulses to precise brain areas via electrodes. If stimulation of an area reliably evokes emotional behaviour, it is likely to be producing the corresponding emotion, particularly if the animal in question confirms its hedonic tone by working to turn off, or receive more of, a stimulus. Panksepp concluded that at least seven emotional circuits are shared by all mammals, including humans, roughly overlapping with our feelings of anger, anxiety, sadness, nurturing, lust, playfulness, and optimistic expectancy. Humans with damage to the structures critical to these circuits hint at their importance: Solms describes a woman with bilateral amygdala damage who lost the ability to feel fear, and as a result has faced unusual dangers, including being ‘held up at both knifepoint and gun-point’.
Emotions have been held variously to be defined by the qualitative experiences which accompany them, the cognitive appraisals they encapsulate, or the bodily changes they trigger. Panksepp’s wholistic model emphasises their neurological basis as well as behavioural outcomes and their evolutionary rationale. Lust exists so we pass on our genes; rage for the protection of resources; fear to escape from danger; care to mediate parent-child attachments. He thought our default state of optimistic expectancy – ‘seeking’, in later articulations, experienced as something close to curiosity — drives us to expand our acquaintance with the world.
In Solms’ account, feelings are not just a foundational aspect of consciousness but also the underlying driver of thinking. Once ‘a need has been prioritised’ as an emotion or affect, ‘thinking generates an expected context in which it can be met.’ Conscious reflection may even originate from the need to unfurl thoughts emotionally. ‘Thinking goes on unconsciously until you’re in a state of uncertainty as to what to do,’ Solms argues. ‘Then you need feeling to feel your way through the problem.’ These arguments give neuroscientific backing to the Humean insight that ‘Reason is, and ought only to be the slave of the passions’.
In the final chapter of The Hidden Spring, Solms outlines his plan to create an artificial consciousness. Already underway, in collaboration with an ‘enlarged team of physicists, computer scientists, biomedical engineers and neuroscientists’, the project will initially simulate, later embody, artificial agents given objectives and constraints: energy requirements and various ways to satisfy them, a ‘thermoregulatory mechanism,’ a ‘pain parameter, triggered by damage to specific aspects of the computer’s physical integrity’, and so on. Next, they will be forced to compete against other similar beings, and the coding of the most successful ‘artificially propagated’ through multiple generations. Solms thinks the winners will have to represent survival-relevant information in their minds in a valanced, qualitative, subjective way, and so the thermal parameter, by way of ‘the precursors of felt affect’, should generate feelings of heat or cold, while anticipation of possible pain triggers might eventually yield felt anxiety. A being’s subjectivity will be inferred when it starts acting in accordance with emotional processing rather than its final aims (as we seek sex, say, regardless of reproduction), at which point Solms intends to shut it down and call a conference.
This project is undergirded by the assumption that the brain’s processes are, in the jargon, ‘multiply realisable’. Solms thinks there’s good evidence now that the functionality of the human brain is duplicable in silico, including the use of radio waves to restore locomotion to monkeys whose spinal cords had been severed, and the creation of a fully functioning, artificial pyramidal neuron. But since a functioning cognitive machine assembled from these neurons must be a long way away (our skulls house roughly 90 billion each), the substrate Solms intends to use is presumably something like a computer, whose ability to realise the brain’s functionality is profoundly uncertain.
One key function of affect as it emerges from this book is to compare and prioritise competing biological needs in different contexts. Couldn’t this middle-layer processing be done by manipulating information alone? Solms thinks emotions are better than algorithms because their qualitatively-distinguished, valanced nature avoids excessive computational complexity. But while it’s true that animal brains make the best of their massively parallel architectures and limited capacities for precise, serial calculation, calculations are precisely what computers are good at: that’s why they beat us at chess. AI research is replete with examples of algorithms cleverly achieving programmed goals by means quite orthogonal to their creators’ intentions.
Solms’ round-about engineering approach is also suggestive. In effect, he proposes arriving at consciousness through evolution, which certainly seems to have worked once. Why go to all the bother, though? ‘As things stand currently,’ he admits, ‘there are too many gaps between the broad brushstrokes I have delineated and the practical implementation of the mechanisms they describe; and these missing steps will only come into view fully as we try to instantiate them.’ The problem is that we can’t characterise the neural basis of affect well enough to design a system that possesses it, and he wants the process to yield the very answers we lack.
Although he suggests compelling reasons for why affective experiences exist, Solms is largely silent about how. The Hidden Spring does contain two abstruse chapters on Friston free energy, which he claims furnish a unifying, mathematical explanation of brain function, but these equations say little about the 700 million year-long process which seems to have turned assemblies of neurons into conscious minds. When Solms suggests we can get ‘surprisingly easily’ from thermodynamics to a feeling being, he’s performing a sleight of hand, and similar leaps across order of magnitude blur the meaning of claims such as that affect is ‘just felt uncertainty’. Partly in consequence, The Hidden Spring can be a baggy book, only half-annealed into a coherent whole, with side arguments such as Solms’ ambition to rehabilitate Freud as the father of any investigation of humans minds distracting from the central lines of inquiry.
Still, it smoulders with provocative and absorbing ideas. An emotional picture of the brain suggests that our default metaphor for the mind — a computer — is even more misleading than it seemed. Rational thinking is something humans effortfully train our emotions to allow us to do. In this regard The Hidden Spring is a valuable adjunct to work that has sought to reinstate the centrality of emotions to cognition, including Antonio Damasio’s Descartes Error and Martha Nussbaum’s Upheavals of Thought: The Intelligence of Emotions.
Solms’ arguments regarding the cortex and the brainstem, meanwhile, though deeply controversial, are certainly testable in the lab, and conclusive evidence either way would have significant implications. Partly because the human cortex is much more developed than that of other organisms, eminent neuroscientists like Vilayanur Ramachandran and Joseph LeDoux think animals do not experience emotions at all. The brainstem is at least 500 million years old, and shared approximately with all vertebrates. If emotional consciousness were generated there, it would imply our planet is teeming with other sentient minds.