What 'high intelligence' actually means here
The phrase is a loose label for a cluster of cognitive capacities that appear together in several taxa: flexible problem-solving, individual recognition of conspecifics, complex social cognition, vocal or behavioral learning that supports cultural transmission, future-oriented planning at modest time scales, and at least some capacity for representing the mental states of others. No single species has all of these capacities at human levels, and the cluster doesn't always travel together — some species have some elements and not others. But the cluster is identifiable enough that comparative cognition researchers can productively use 'high intelligence' as a category even while debating its boundaries.
Primates (including humans, great apes, and some monkey species) develop high intelligence via a large neocortex with detailed mammalian features.
The convergent lineages
Primates (including humans, great apes, and some monkey species) develop high intelligence via a large neocortex with detailed mammalian features. Cetaceans (dolphins, whales) develop high intelligence via large brains with mammalian neocortex evolved for aquatic life. Elephants develop high intelligence via similarly large brains with mammalian neocortex evolved for their ecology. Corvids and parrots develop high intelligence via large avian forebrain (pallium) with columnar organization but no mammalian-style six-layer cortex. Cephalopods (octopus, cuttlefish, some squid) develop high intelligence via a fundamentally different neural architecture distributed across central brain and peripheral nerve cords, with no cortex of any kind. The neural substrates are distinct; the cognitive capacities are convergent.
What convergence implies
Convergent evolution of similar traits in distantly-related lineages typically indicates that the trait is favored by similar selection pressures and is achievable via multiple independent biological pathways. The convergent emergence of high intelligence across these distant lineages suggests two things. First, intelligence is selectively favored under certain ecological conditions — complex social environments, varied food resources requiring flexible problem-solving, life-history strategies that involve long-term investment in offspring or social relationships. Second, intelligence can be supported by multiple neural architectures rather than requiring any specific brain plan — corvid pallium, cephalopod nerve-cord distribution, mammalian cortex all produce sophisticated cognition in their respective lineages. The traits are convergent because the selection pressure is universal; the neural substrates are divergent because evolution finds multiple solutions.
What this means for the singular-human-intelligence framing
The convergent-evolution data is hard to reconcile with the older framing that human intelligence is a unique evolutionary breakthrough. The cognitive capacity cluster appears repeatedly in evolution, in distantly-related lineages, via multiple neural architectures. Humans have an unusual elaboration of the capacity cluster — particularly in language, abstract reasoning, and cumulative cultural elaboration — but the underlying cognitive machinery is not species-unique. This reframing changes how comparative cognition research approaches the question of what makes humans cognitively distinctive: not the presence of certain capacities, but the particular degrees and combinations of capacities that elaborate beyond non-human levels.
What this means for crow research specifically
American crow cognition fits the convergent-intelligence pattern. The species has flexible problem-solving, individual recognition (Marzluff[2] face-recognition work), social cognition (mobbing recruitment, social information transfer), vocal learning (the substrate for any communication-system complexity), and at least some future-oriented planning capacities. The cognitive profile is comparable in many ways to that of other high-intelligence lineages. The neural substrate (avian pallium with cortex-like columnar organization, as established by the 2020 Stacho work) supports the cognitive capacity at neuroanatomical levels analogous to mammalian cortex. The interpretation that crows are 'smart' isn't projection onto a species — it's accurate description of the cognitive profile the species shows, with that profile fitting the broader convergent-intelligence pattern across distant lineages.
What this means for animal-communication AI research
The convergent-intelligence pattern provides framing for what to expect from AI bioacoustic research on different species. Species with high-intelligence cognitive profiles — corvids, cetaceans, elephants, possibly cephalopods — have the underlying cognitive substrate that could support more-than-reflexive communicative content. Whether they actually have communication systems that exploit that substrate richly is an empirical question, but it's at least one that's biologically plausible. Species with simpler cognitive profiles have less plausible candidate communication systems for the rich-structure framing. The convergent-intelligence framing is one of the reasons that contemporary AI bioacoustic research focuses on the cognitive-elite species rather than on broader taxonomic samples. It's also why the realistic expectations for these projects involve substantial findings rather than easy translation — the species are likely doing communicatively interesting things, but the systems are probably not structured like human language regardless.