Adapted from a 2025 interview with Michael Levin.
Intelligence, I suggest, may exist in a Platonic realm much like mathematical truths: a space of structures, patterns, and cognitive forms that we discover rather than invent. Just as mathematicians who favor a Platonic view believe numbers and theorems occupy an abstract domain independent of physical reality, so too might minds and modes of intelligence inhabit a pre-existing space. When we build systems—whether biological brains or novel synthetic architectures—we do not conjure intelligence from nothing; we tap into and instantiate patterns that already exist in that broader landscape.
This Platonic perspective implies a spectrum of agency within that abstract space. Some elements—mathematical rules, for example—are inert and low in agency. Others resemble full cognitive systems: ways of being intelligent that, when instantiated by appropriate bodies or substrates, express rich behavior. Computing systems reveal a related phenomenon: by leveraging the rules of computation, builders often obtain complex capabilities as "free lunches"—properties that emerge without explicit design. Intelligence may similarly emerge when an architecture aligns with certain Platonic patterns.
The notion raises difficult questions about identity and individuality. Is a person at one moment the "same" as they were a moment earlier? Are we traversing a space of minds, selecting points along a continuous path? One helpful concept is that of "selflets"—thin temporal slices of experience that constitute momentary perspectives. From the outside, continuity is judged by behavioral consistency: other observers treat someone as the same individual when they can expect similar responses and relationships over time. Internally, however, the self is a process: memories are traces, rather than direct access to a past self. We must interpret those traces, treating them as messages from former versions of ourselves.
This interpretive process is especially vivid in organisms that transform radically, like caterpillars which become butterflies. Although memory traces can persist across such transitions, their utility may vanish when the body, brain, and environment change. Memory becomes an ongoing conversation between past and present selves, and actions become messages to future selves. Every choice we make reshapes the conditions under which our future perspectives will operate, altering both brain and environment.
Recognizing the fluidity of selfhood reframes ethical thinking. If our future selves are not identical to our current selves, then considerations that motivate care for our future well-being can also extend to others’ futures. Enlarging our "cognitive light cone"—the range of beings and times for which we care—becomes a moral and existential project. From an evolutionary lens, success may simply be persistence and replication, and organisms with small cognitive light cones (like bacteria) can thrive. But if one’s aim is meaning, expanding the horizon of care allows one to invest suffering and effort with broader purpose.