Constant and seemingly random movement of fish serves to optimise their sensory feedback, which may enhance the way sensors are deployed in robot design
One characteristic of any large group of animals IS they seem to move constantly and almost at random. It is particularly pronounced when observing a large number of fish; they flicker and wriggle in a constant blur of sparkling scales. Researchers at Johns Hopkins University have found that these movements are in fact not random at all, but serve to maximise what the fish can extract from their environment, a finding that the researchers believe may have profound implications for robot design.

Mechanical engineer and roboticist Noah Cowan, who led the study, explains that it was already known that all animals have to actively move in order to sense their environment. “There’s a saying in biology that when the world is still, you stop being able to sense it,” he said. To test how this worked, he and his colleagues, including researchers from the New Jersey Institute of Technology (NJIT), studied fish that generate a weak electric field around their bodies, which is believed to help them sense their environment and communicate with each other.
The study involved watching how the fish moved inside a transparent tube submerged in a tank. When the tube was kept still, the fish constantly wriggled back and forth. To challenge the fishes’ senses, the team moved the tube in two ways: synchronised with and opposite to the animals’ movement. When the tube’s movement gave them less sensory feedback – that is, when it was moving synchronised with them – they swam farther, and when moving opposite to them they swam less. In the dark, when the fishes’ electric senses were the only way they could receive any feedback on their environment, their movements were even more pronounced.
“What we found that wasn’t known before is animals constantly regulate these movements to optimise sensory input,” said Cowan. “The action to perceive the world is under constant regulation,” added Eric Fortune from NJIT. “We think that’s also true for humans.” One example of how humans use movement to optimise sensory input is moving our hands up and down to estimate the weight of an object we are holding, the team speculates in a paper in Current Biology describing their research.
The lead author on the study, Debojyoti Biswas, notes that engineers do not take account of the importance of active movement to optimise sensory input when designing the placement of sensors on robots, but understanding how this can improve information-gathering about the environment could help improve robot design. “Knowing more about how these tiny movements work might offer new design strategies for our smart devices to sense the world,” he said.
I suppose the human parallel is the ‘group’ manner in which a particular animal (us!) behaves at sporting events and popular music festivals. The concept(s) of swarm behaviour (in birds, insects and fish) surely offer even more insight. What a fine piece of deduction by staff at NJIT.
Speaking as a fairly short person who attends music festivals, my behaviour to optimise sensory input usually involves bobbing up on tiptoes and craning around to one side or the other to see round the inevitable tall person in front of me.
This is interesting. Does it mean the sensors (temperature, load cell, etc.) on my experiment will work better if I install a small variable frequency vibration to the test stand? Hmmm….
We are actually estimating inertia when we move an object up and down slightly to judge how “heavy” it is. In other word we are measuring inertial mass, which is likely to be the most significant thing to us in practically utilizing the object which is usually in a dynamic way.
I’m not sure that the benefits of movement will be as great for robots as predicted. I supect that one of the reasons for animal movement to improve sensory perception is that the output form most biological sensors fades with time unless the stimulation changes. The animal movement will act to change the stimulation levels of its senses and keep everything “switched on”. The detectors used in robotic systems don’t need this re-stimulation to deliver a continuous output.
Great point — and that’s one of the things we have thought about; there are some really neat new emerging sensors, like cameras that only report change of their pixels, that can really enhance the dynamic range of sensors, but that would require this type of active sensing.
Interesting – “bio-like” behaviour enabling the use of “bio-like” sensors. I guess additional benefits of this type of sensor will be reduced power consumption and reduced bandwidth needed to deleiver their information.
We are all, for the most part, wonderful examples of Engineering and applied technology: and of course have the added bonus of being able to ‘re-furbish’ most of the stressed parts as we go along! At least for a time. [78 and still counting, I believe is getting towards the limit!]
My dentist assures me it is the stimulation of the minor movements they receive in the action of chewing which has kept my teeth and jaw-bones ‘sound’: likewise the constant use and exercise in movement of limbs keeps their ‘opposite’ muscle and joints up to scratch. Perhaps fish worked this out early in the ‘cycle’ of evolution such that they are so successful.
I recall my biology teacher telling us (O level! GCSE for younger readers) that the dimension of the human eye-socket evolved and was linked to the size of the human hand, when ‘clenched’ around the most likely ‘object’ a stone which would easily fit into the hand before throwing , to be used as a projectile against another human!-perhaps the most ‘dynamic’ manner possible!