Somewhere on Alaska’s North Slope in the spring of 2025, a grizzly bear picked up a rock, adjusted it in her paws, and brought it down on a salmon carcass. The moment lasted seconds. But a small camera bolted to the bear’s collar caught every frame, and when researchers at Washington State University reviewed the clip months later, they found themselves watching something that had never been documented in a wild brown bear: deliberate tool use to process food.
The footage is part of a massive, still-unclassified dataset from a WSU-led collar-camera project that has generated more than 400 hours of first-person bear video from one of the most remote landscapes in North America. Lead researchers Ellery Vincent and Jordan Pruszenski collared roughly 12 of an estimated 200 grizzlies on the North Slope, fitting each animal with two camera collars that record 4-to-6-second clips at set intervals, capturing up to about 17 hours of footage per collar. The result is an unprecedented visual archive of daily grizzly life: foraging, digging, social encounters, and now, possibly, toolmaking.
Why this would matter
Brown bears are not supposed to do this. Among large carnivores, tool use is vanishingly rare in the wild. The only peer-reviewed record of a wild brown bear using a tool comes from a 2012 paper in the journal Animal Cognition, accessible through a Springer-hosted DOI. In that study, a bear in Southeast Alaska repeatedly picked up barnacle-encrusted rocks and used them to scratch its own face and neck. The behavior met the formal scientific definition of tool use: selecting, manipulating, and applying an external object to achieve a goal.
But scratching an itch and cracking open a skull sit on very different rungs of the cognitive ladder. Using a rock to process food implies planning, material selection, and goal-directed force. If the North Slope footage holds up under formal analysis, it would represent a significant expansion of what scientists understand about grizzly intelligence and problem-solving.
What the collar cameras are revealing
Collar-camera technology is not new to Alaska. The Alaska Department of Fish and Game has been distributing collar-cam footage of bears for well over a decade, using the technology alongside GPS tracking and physical tagging as a standard wildlife management tool. What has changed is scale. Smaller cameras, cheaper batteries, and larger digital storage have made it possible to outfit multiple bears simultaneously and collect volumes of data that would have been financially impossible ten years ago.
The WSU project exploits that shift. Traditional bear research relies on GPS tracks, scat analysis, and occasional distant observation, all of which force scientists to infer behavior from indirect clues. A collar camera records exactly what a bear sees, touches, and eats. Behaviors that might have gone unnoticed for decades can now be cataloged, replayed, and reviewed independently by multiple researchers.
A post-den emergence encounter recorded in May 2025 offered some of the project’s earliest behavioral observations, including foraging sequences, exploratory digging, and social interactions between bears. But the sheer volume of footage means classification is painstaking. Each clip must be tagged and categorized frame by frame before any behavioral pattern can be confirmed as statistically meaningful rather than a one-off curiosity.
What has not been proven yet
As of June 2026, no published, peer-reviewed study has described a wild brown bear fashioning or selecting tools specifically to break open salmon skulls. The WSU team has not yet released classified results from the collar-cam footage, and no timeline for formal publication has been announced. The salmon-skull observation, while captured on video, remains preliminary until it survives the peer-review process.
That distinction is not a technicality. Anecdotal reports of bears using rocks or logs to manipulate food have circulated among Alaskan guides and fisheries workers for years, but anecdotes do not meet the standard of peer review. They are often based on brief, high-adrenaline encounters with no video or systematic notes. The collar-cam footage could eventually confirm or refute those stories, but until the classification work is complete, even detailed field accounts should be treated as hypotheses.
Captive studies add some supporting context. Research on sun bears in controlled environments has shown that at least some bear species can learn to use tools when presented with novel problems. But captive behavior does not automatically predict wild behavior, and no captive study has documented the specific sequence of selecting a rock and applying it to a fish carcass.
Where the science goes from here
Two converging trends make this a pivotal moment for bear research. The hardware problem has largely been solved: animal-borne cameras are now small and durable enough to survive months on a grizzly’s neck in Arctic conditions. And advances in machine learning are beginning to make it feasible to sift through hundreds of hours of short clips, flagging sequences that show rare behaviors like tool use, cooperative hunting, or interactions with human infrastructure.
If repeated instances of tool use for feeding are documented and verified across multiple bears, scientists may need to revise long-standing assumptions about how grizzlies solve problems, adapt to shifting food supplies, and navigate landscapes increasingly shaped by human activity. Even if the salmon-skull observation turns out to be a single, unrepeated event, the collar-cam projects now running on the North Slope ensure that future claims can be evaluated against a rich visual archive rather than fragmentary stories told around campfires.
For now, the most honest reading of the evidence is this: wild brown bears have been conclusively shown to use tools for self-grooming. Collar cameras are recording grizzly behavior in detail no previous technology could match. And somewhere in more than 400 hours of unclassified footage, there may be proof that at least one bear on Alaska’s North Slope figured out how to crack a salmon skull with a rock. The clip exists. The peer review does not, not yet.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
