Problem-solving

In comparative cognition, problem-solving usually refers to an animal reaching a goal — often food, escape, or access — when the direct route is blocked and some novel or flexible behaviour is required. Researchers are interested in whether the animal can find a solution it has not simply been trained to perform, and how it gets there: by trial and error, by learning from others, by reorganising familiar actions, or in some debated cases by appearing to work out a relationship before acting.

A key distinction runs through the whole field: observable behaviour versus inferred internal states. We can record that an octopus removed a screw-top lid, or that a crow dropped stones into a tube to raise a water level. What we cannot read directly is what the animal understood while doing it. Careful studies try to separate associative learning (gradual reinforcement of what works) from more flexible cognition, but the boundary is genuinely difficult, and many behaviours involve elements of both rather than a clean either-or.

Because of this, researchers tend to describe results in cautious language — "is consistent with", "may indicate", "in this studied population" — rather than declaring that an animal "understands" or "thinks". That caution is not vagueness; it reflects how hard it is to rule out simpler explanations for an impressive-looking result.

Some members of the crow family — including certain populations of New Caledonian crows (Corvus moneduloides) and some rooks and ravens in captivity — have been studied solving multi-step physical tasks. Documented examples include using and shaping stick-like tools to extract food, and dropping stones into a water-filled tube to bring a floating reward within reach. These are real, repeatable findings from controlled studies, not viral clips, and they are striking because tool use of this kind is uncommon among birds.

It is important not to overgeneralise from them. Tool-related behaviour is well documented in particular corvid species and populations; it is not a property of "all crows" or of birds in general. Some of the most cited stick-tool behaviour is associated with wild New Caledonian crows specifically, where it appears in natural foraging, while other corvids show flexible problem-solving mainly in experimental settings. Whether a given solution reflects insight, prior experience, or step-by-step learning is often still discussed, so the safest summary is that these species show flexible, sometimes tool-assisted problem-solving under the conditions studied.

Great apes such as chimpanzees (Pan troglodytes) and orangutans (Pongo species) have a long history in cognition research, including using objects as tools, combining actions to reach rewards, and learning from watching others. Wild chimpanzee populations have been observed using sticks and stones in foraging, with local differences between communities. These descriptions stay close to what is observed; claims about apes "reasoning like people" go well beyond the evidence and are avoided in careful work.

Octopuses (class Cephalopoda) are a striking contrast — invertebrates with a very different nervous system, much of it distributed through the arms. In captivity some individuals have been reported opening containers, navigating mazes, and manipulating objects, and there is research interest in their flexible behaviour. Findings vary between individuals and studies, and much octopus work involves small numbers of captive animals, so conclusions are held loosely. Elephants (family Elephantidae) have likewise been studied on cooperative and physical tasks, and have been a focus of mirror-mark studies in a few individuals; results are intriguing but limited in sample size and should not be stretched into broad claims about the whole group.

The wider lesson is that intelligence is not one thing measured on one scale. A corvid, an ape, an octopus, and an elephant inhabit different bodies, senses, and ecological pressures. Comparing them on a single "who is smarter" axis ignores that each is solving the problems its own way of life presents, which is why researchers describe cognition as context-specific rather than ranked.

How a problem is presented can change the outcome as much as the animal's ability. A task that suits one species' body, senses, and motivation may be nearly impossible for another for reasons that have nothing to do with cognition. A test relying on visual cues disadvantages an animal that leads with smell or touch; a food reward an animal does not value, an apparatus it cannot grip, or an unfamiliar testing room can all depress performance. Researchers spend considerable effort designing tasks that are fair to the species being tested, and re-examining tasks where the design may have favoured one group.

This is why the field treats cross-species comparisons with caution. If two species are tested on the same apparatus, the one whose body and senses suit the apparatus has an advantage built into the test itself. Differences in results may reflect the task as much as the mind behind it. Good comparative studies try to control for these factors, report their methods openly, and avoid sweeping claims — an approach consistent with how FaunaHub handles sourcing through its animal research sources methodology.

A central principle in this research is that a negative result is weak evidence. If an animal does not solve a problem, the reason might be the design — wrong reward, wrong sensory channel, stress, lack of motivation, or simply that the individual tried a different strategy. It does not establish that the species cannot solve such problems, only that these animals did not, under these conditions. This is the mirror image of the rule for positive results, where researchers work hard to rule out simpler explanations before crediting an impressive performance.

The same caution applies to widely discussed tests such as the mirror self-recognition test, where an animal is checked for whether it reacts to a mark visible only in a mirror. Passing it is not proof of human-like consciousness, and failing it is not proof that an animal lacks any sense of self. The test has known limitations and is biased toward species that rely heavily on vision and naturally attend to mirrors, so results are best read narrowly rather than as a verdict on awareness.

Where an animal is studied matters. Captive settings allow controlled, repeatable tasks but can also encourage behaviours rarely seen in nature, shaped by training, boredom, or the apparatus itself. Wild observations capture problem-solving in its natural context but are harder to control and interpret. A behaviour reliably produced by captive individuals should not be assumed typical of wild populations, and vice versa; the two lines of evidence inform each other but are not interchangeable. Enrichment in captivity is sometimes part of how these behaviours are studied, and is mentioned here only as research context, not as advice.

Finally, problem-solving sits next to questions about emotion and communication, and the same discipline applies. Stress, persistence, or apparent frustration during a task are described from observable behaviour, not asserted as definite inner feelings. And solving a problem, or signalling about one, is communication or cognition — it is not evidence of human-like language or of a mind that works like ours. Keeping observation and inference separate is what allows this field to say something meaningful about animal problem-solving without overclaiming.

How the claims on this topic are studied and read:

• How animal intelligence is studiedField observation, controlled tasks, and comparative cognition — and why one task never defines a whole mind.
• Tool-use definitionsWhat counts as tool use, the contested boundary cases, and why ‘technology’ is the wrong frame.

How whole groups of animals show this behavior:

• Corvid intelligence
• Cephalopod intelligence

This guide is part of FaunaHub's animal intelligence & behavior cluster. For how these claims are sourced, see animal research sources, and for the biology behind behavior, see animal senses & adaptations.