Starlings can tell if you are watching them, according to a study that has shown for the first time that starlings respond to a human's gaze.

Starlings will keep away from their food dish if a human is looking at it. However, if the person is just as close, but their eyes are turned away, the birds resumed feeding earlier and consumed more food overall, according to experiments by Julia Carter and colleagues at the University of Bristol, reported today in the Proceedings of the Royal Society Biological Sciences.

Could this be exploited to make a better scarecrow? "Starlings do seem to have a reasonably strong aversion to eyes, even to artificial eyes, but these birds are also very quick learners," she says.

"Previous studies have shown that starlings will learn within a matter of hours to ignore even relatively elaborate bird scarers - these devices never do what a real predator would, they don't actually chase the birds or present any other signs of danger, so the birds quickly learn to ignore them."

This fear of being observed directly may be hard wired into bird brains, since predators tend to look at their prey when they attack, so direct eye-gaze can predict imminent danger. Equally, it might be a sign that starlings are doing something smarter than that, by taking into account the view of another, not just another bird but another species.

Carter says:"It is not yet clear what this means for the intelligence of starlings, they are certainly far more observant and sensitive to human facial cues that we might have imagined - what we find is responses to really very subtle differences in a human's eye-gaze, even though there are far more conspicuous and salient cues that don't change in the experiment: the close proximity of the person, and their body and face orientation."

"The possibility that the starlings can "read the intention of another species" is certainly an intriguing one, and is indeed an interesting possibility, though there are likely alternatives to consider," she says.

"The simplest explanation might be that these birds are responding at an innate instinctive level: circular eye-like patterns are thought to be rather conspicuous to vertebrates like starlings, so a human who is staring directly at the starling is likely to attract the bird's attention (because the direct eye-gaze presents a completely visible circular pattern) more so than the averted eye-gaze (which presents an angled view, and therefore a less circular pattern)."

Another likely possibility is that the starlings have learned to fear a direct look. "If a starling is ever chased or captured in the wild or in captivity, the event is preceded by the captor looking directly at the bird.

"A starling is never chased if it is not looked at first, so a predictive relationship can be learned - looking leads to a risk of capture. The final possibility is that the starling can recognise that the human is looking at it, and infer the likelihood that they will consequently be chased."

But, she says, "It is notoriously difficult to get inside the black box of an animal's brain in order to establish, for example, whether they can appreciate another's visual perspective or even whether they have a 'theory of mind'.

The reason that they benefit from being sensitive to a human gaze is down to peer pressure. Wild starlings are highly social and will quickly join others at a productive foraging patch. This leads to tough competition for food so an individual starling that assesses a relatively low predation risk, and responds by returning more quickly to a foraging patch (as in this study), will gain valuable feeding time before others join the patch.

"By responding to these subtle eye-gaze cues, starlings would gain a competitive advantage over individuals that are not so observant. This work highlights the importance of considering even very subtle signals that might be used in an animal's decision-making process."

Responses to obvious indicators of risk - a predator looming overhead or the fleeing of other animals - are well documented, but Carter argued that a predator's head and eye-gaze direction are also useful indicators of risk, even though subtle, since many predators orient their head and eyes towards their prey as they attack.

Carter did the study with Nicholas Lyons, Hannah Cole, and Arthur Goldsmith.