The experimental design
Swift's funeral-behavior experiments staged human-controlled scenarios in which a person, sometimes wearing a distinct mask, presented to a group of crows at a known feeding site while holding a dead crow. The presentations were repeated, contrasted with controls (the same person presenting without a dead crow, or presenting a non-crow dead animal), and the crow responses were systematically recorded. Crows responded to the dead-crow presentations with intense mobbing-like vocalizations, gathering, persistent attention to the holder, and subsequently treating the holder as a threat across future encounters. The behavior was real, repeatable, and species-typical.
The findings show that crows treat dead conspecifics as significant events that elicit specific group-level behavioral and vocal responses.
What it shows and doesn't show
The findings show that crows treat dead conspecifics as significant events that elicit specific group-level behavioral and vocal responses. The findings show that the human associated with the dead crow becomes a learned-threat target in the crows' subsequent behavior. The findings don't show that crows are 'mourning' in the human emotional sense — the behavior pattern is consistent with several non-mourning interpretations including learning-from-conspecific-death (acquiring information about predator types), social information transfer (alerting others to a localized danger), or threat-assessment behavior (treating dead-conspecific as a marker of nearby risk). The 'funeral' framing is human projection onto behavior that has clearer functional interpretations.
Why the methodology matters
Swift's work is methodologically careful in ways that distinguish it from anecdotal 'crows held a funeral' observations. The controls rule out the possibility that any group gathering at a feeding site looks funeral-like. The repeated presentations rule out one-off coincidence. The mask manipulation tests show that the response is specific to the human holding the dead crow rather than humans in general. The cross-presentation tests show that the response extends to non-conspecific dead animals but at lower intensity. The methodology produces conclusions that survive scrutiny — which is why the work was publishable in peer-reviewed venues (Animal Behaviour 2015, with subsequent follow-up papers) rather than just blog posts.
Brain-imaging follow-up
Swift and Marzluff[1] followed up with PET brain-imaging studies on crows responding to threatening human faces (including those associated with dead-conspecific presentations). Active brain regions during the threat response include areas analogous to mammalian fear-processing and learning circuits, providing neural correlates for the behavioral findings. The neuroscience extends the behavioral work into a richer characterization of what's happening cognitively when a crow encounters a dead conspecific and the human associated with it. This is exactly the kind of converging behavioral-plus-neural evidence that supports robust interpretation rather than over-extension.
How the public coverage went wrong
Press coverage of the funeral work mostly emphasized the 'crows hold funerals' framing, which the research itself doesn't directly support. The behavioral pattern is real; the cognitive interpretation of mourning is human projection. Swift herself has written extensively in public communication (her Corvid Research blog runs since 2014) clarifying what the research does and doesn't support, and pushing back on the more sensationalized framings. The gap between rigorous behavioral research and public communication of it is a recurrent theme in corvid science, and Swift's case is one of the cleaner examples of a careful researcher trying to keep the public framing honest.
Why this matters for CrowLingo
When crows gather and vocalize intensely around a dead conspecific, the atlas would categorize the calls as some combination of mobbing and assembly behavior. The behavioral-context cluster the calls fall into is real and species-typical. The cognitive interpretation of what the crows are 'feeling' is a layer above what the bioacoustic data supports, and the atlas treats that distinction carefully — by labeling the behavioral context (mobbing/assembly) rather than the emotional interpretation (mourning/grief). The Swift research is one of the clearest demonstrations of why the careful framing matters.