Everybody loves a good debate about the mystery of consciousness. Physicists get to invent quantum mechanical explanations of macroscopic phenomena. Humanities folk get to poo-poo the evils of reductionism. Journalists can trot out their full repertoire of both scientific and ethical cliches. Everybody gets to argue. There are good working theories of consciousness that are much more compelling than quantum mechanical explanations or folk theories, but they are neither fully satisfying nor at all established.
The thing is, when it comes to decision-making, they don’t matter that much. The key lessons for a decision-maker from cognitive science aren’t dependent on theories of consciousness. They are baked so thoroughly and deeply into the structure of our brains that whatever story about consciousness ends up being the right one, it isn’t going to change any of the key facts about cognition and transformative experience.
We learn from experience and learning changes who we are at the most basic level. Experience creates memories, of course. But it does far more than that. It changes the way our minds are connected, and the way our minds are connected is, quite literally, the way we think. The way we think is definitional of who we are.
Modern cognitive science shows us to be learning machines. We are adaptive learners whose basic cognitive architecture ensures that experience will change us. We can think about and to some extent control how we change by choosing different experiences. The foundational problem of decision-making beyond preference optimization – the need to select between experiences that will change our preferences – is a direct result of our cognitive architecture.
Though connection systems are far from intuitive, the implications of a connection architecture fit very well with our internal experience of thought and the behavior of people in general. This feels like who we are and how we think. Our cognitive toolkit creates and limits the pathways we have for making self-altering decisions. It’s hard to eat soup with a fork or cut bread with a spoon. For someone facing a transformational choice, cognitive science provides a useful description of how cognitive structure is formed, the limitations we face in making decisions, and the biggest risks to rational choice.
The massive prediction capabilities we enjoy and the focus of cognition on action should make us very well equipped to make preference-changing decisions. After all, such decisions are all about prediction and action. That said, our cognitive toolkit is far from perfect. Connection systems are amazingly capable learning devices. But like any learning device, they are dependent on the experiences they have. In the realm of embodied motion and sensory perception, this isn’t an issue. We’re awash in a vast sea of constant experience – so our brains become enormously adept at deciphering and navigating our complex physical world. That isn’t the case with a great many other types of social learning. There are countless situations where we must rely on one or a very small number of experiences. That’s problematic for our kind of learning and it results in all sorts of sub-optimal cognitive behaviors.
Our brains do not embody anything like perfect rationality. We must learn not to overweight initial experiences. We must find strategies to avoid the cognitive hardening and lack of intellectual flexibility that comes from repeated reinforcement. We need to learn how to compensate for experiences we wish we hadn’t had, and we must work to find experiences that will not change us irrevocably in ways we do not desire.
The fact that connection systems don’t naturally deliver something closer to perfect rationality shouldn’t be all that disturbing. After all, rationality isn’t the point of life, it’s a tool for better decision-making. And decision-making involves a great deal more than classical rationality – much of which is better described by learning. Our brains are incredible learning systems, so flexible that we can even learn to better approximate perfect rationality. In fact, the natural tendencies of human cognition toward sub-optimal rationality are far less damaging to the most important decisions in our life than some other characteristics of connection-system architectures.
We struggle with the black-box nature of connection systems. A complex connection system can’t be translated into words or structured procedures. This creates a barrier between us and others that can never be completely removed. It’s a barrier that is fundamental to being human and lives at the heart of our most important emotional experiences. Loneliness. Friendship. Love. The inability – unavoidable and unfixable – of language to accurately convey thought defines the human condition and limits the extent to which we can learn from each other’s experience.
Connectionist thinking doesn’t just limit communication between people. It puts hard limits on our ability (sorry Polonius!) to know ourselves. The same limits that apply to communication plague introspection. We really don’t know much about how we think or why we think what we do. The world blithely assumes that while it may be ignorant of our character, we – at least – must know it. Not so. There is a very real sense in which our own character and thought are, and will always be, mysterious even to ourselves.
Even more troubling, at least from a practical decision-making perspective, our minds do not provide imaginative access to how we’ll react to any experience. Though we are quite good at abstract prediction and capable of re-litigating experience in countless ways, we are miserable at projecting how an experience will change us or what an experience will be like.
Our dependence on experience to shape and expose thought creates a more abstract, philosophical concern. We use experience and prediction to bootstrap from almost nothing to complex explanation and prediction of the world. The experiences we have are, of course, a by-product of the culture we exist in. Humans are social animals and most of what we care about involves others. Most of what we think comes from others. Most of how we feel is driven by others. It would be nearly impossible to overstate the degree to which we are shaped by and concerned with other people.
Which raises a question about the extent to which our thinking can ever transcend our local culture. When the very mechanisms of our thought are socially constructed, how much confidence can we have in the genealogy of even our most considered beliefs?
It’s hard to know, but we are not completely adrift in a relativist sea. Cognition is embodied. Its initial systems are driven by basic biological desires and needs focused on effective action and anchored by the real world. Many of our core cognitive capabilities are created in experiencing and acting in what we like to think of as reality. While still in the crib, we learn foundational lessons about cause and effect, about the divide between the self and other, and the difference between people and things. There is reason to think that the sensory world and our embodiment provide forms of learning and cognitive development that we have good grounds to trust and that can usefully serve as anchor points for thought that lives outside of – and can be used to reflect on – a local culture.
For philosophers, the body has usually been the problem vexing cognition: the seat of desire, emotion and illusion that clouds an otherwise pristine rational clarity. How ironic that the truth may be just the opposite. We likely owe to our embodiment whatever grounds we have for trusting our cognitive toolkit.
All of this gives us grounds for both hope and concern. Modern neuroscience suggests that human cognition is neither iron-clad nor hopelessly flawed.
What a surprise.
Philosophers like to say that you can’t derive “ought” from “is” and that’s mostly true. Nothing in modern neuroscience can tell us what decisions are right or even how to make good decisions outside of preference optimization. Understanding cognition doesn’t magically bridge the gap between the classic model of rational decision-making and transformational choice. It does make clear why that gap is THE problem for thinkers like us. Why it’s so hard to tackle. And even, to some extent, what kind of answer might work.
It’s possible to imagine an intelligent creature that wasn’t an adaptive learner but was simply born with a pre-wired and essentially static intelligence. Or a creature might be so constituted that it could communicate the state of its mind effortlessly and fully understand its cohabitants. It’s possible that our thinking could have developed so that we could simply tweak mental code directly and be a different person. “Ah yes – line 43,000 of this procedure in V1 should really be “Less than or equal” not “less than”. Changed! It’s possible that a thinker could evaluate experiences and know exactly how having them would change its own thought. To such a creature having the experience wouldn’t matter. Thought would be as good as reality.
But while it’s possible to imagine these different kinds of intelligence, the more we learn about cognition, the less likely it is that anything intelligent could do these things. And whether those kinds of intelligence are possible or not, we aren’t that kind of thinker. Nor are we, as philosophers like to claim, a rational animal. We are an animal that must work HARD to be rational. The difference is profound.
We live in a world where the biggest and most important decisions we make come with deep and irrevocable cognitive uncertainty.
This is what it means to be human.