Elephants have long been credited with intelligence, but the claim has often rested on anecdote - an elephant that seems to remember a keeper after years, a working elephant that refuses to pull down a marked tree. Systematic research over the past three decades has put much of this folklore on firmer ground, and it has also complicated it. Elephants do perform remarkably on certain cognitive tasks and unremarkably on others, and their cognition is best understood on its own terms rather than as a version of human or primate thinking.

Elephants recognise themselves in mirrors, a test that only a small number of species pass and that is often taken as evidence of self-awareness. The test in its standard form places a visible mark on the animal's body, in a location the animal cannot see directly, and then observes whether the animal uses the mirror to investigate the mark. Asian elephants studied at the Bronx Zoo in the 2000s passed this test, behaving in ways consistent with recognising the image as themselves rather than as another individual. The study, led by Joshua Plotnik, has since been replicated with modest sample sizes at other facilities. Self-recognition alone does not establish a rich inner life, but it is one of several pieces of evidence that suggests elephants occupy a cognitive space shared with great apes, dolphins and a few corvid species.

Social intelligence is perhaps the most thoroughly studied aspect of elephant cognition. Field research on African elephants by Cynthia Moss, Joyce Poole and their colleagues over several decades has shown that elephant herds are organised around matriarchs, older females whose accumulated knowledge includes the locations of distant water holes, routes used only during droughts, and the recognition of other elephant groups. Herds led by older, more experienced matriarchs survive droughts better than those led by younger ones, a pattern documented in Amboseli National Park over many years. Ethologist Dr. Kofi Addo has argued that this suggests an ecological role for specific individuals, whose loss - for example through poaching of older females for their larger tusks - damages the herd's prospects in ways that the population figures alone do not capture.

Memory is a particular strength. Elephants recognise the calls and smells of other individuals after years of separation, and they respond differently to calls of close relatives than to calls of unrelated elephants, even when the calls have been recorded and played back in experimental settings. They also appear to remember the locations of specific resources over very long periods. The old phrase about 'elephantine memory' has, in other words, a factual basis, though it refers to specific memory systems rather than a general ability to remember everything.

Tool use is less advanced. Elephants do use tools - they pick up branches to scratch themselves in places their trunks cannot reach, and they strip leaves from twigs for use as fly whisks - but their tool use is limited compared with that of chimpanzees or some crows. The structure of the trunk itself, which is extraordinarily dexterous, may reduce the selection pressure for external tools: an elephant can do many things with its trunk that other large animals would need implements for.

Cooperation experiments have produced intriguing results. Plotnik and colleagues designed a task in which two elephants had to pull simultaneously on ends of a rope to reach food. Successful elephants waited for their partner to arrive before pulling, and in some cases returned to the experimental area with a partner who had wandered off. Similar experiments in primates have required extensive training; elephants in these studies understood the task quickly, suggesting a social template for cooperative behaviour that their species possesses.

Grief has been widely reported, but the question of what exactly elephants feel around the death of companions is difficult. Elephants do show striking behaviour around dead members of their species - repeated visits to the remains, prolonged sniffing and touching with the trunk, slow movement of the body - and these behaviours can continue for days or weeks. Whether these constitute grief in any sense comparable to human grief is a question that empirical research can only partly address. Dr. Addo has argued for a careful middle position: elephant behaviour around death is clearly different from their behaviour around other stressors, and the emotional implications, whatever their precise character, deserve recognition in welfare and conservation policy.

The practical implications for conservation are considerable. Elephants are not interchangeable units in their herds; specific individuals carry knowledge that their group depends on. Enforcement against poaching, protection of habitat, and the maintenance of corridors that allow herds to move between resources all take on different meanings when the cognitive dimension is taken seriously. Dr. Addo has argued that elephants are 'ecologically cognitive animals': their effect on their environment depends not only on their physical presence but on what they know. Protecting them, therefore, means protecting their ability to learn, remember and pass on knowledge across generations.