There are approximately 30,000 species of fish and within all these species there is an impressive range of behavioural adaptations. Fish presents an interesting group for the study of animal learning as, at first glance, the fish brain appears simpler than those of other vertebrates, especially primates. The fish brain is proportionally smaller than that of other vertebrates (e.g. birds, mammals and primates) and lacks a neocortex. The neocortex is the section of the brain that most people would recognize as the folded bit on top. This brain section is where most of our advanced abilities such as counting, abstract learning and language occurs. Because fish lack the neocortex, their brain is sometimes called ‘simple’.
While the fish brain may be smaller and ‘simpler’, we know they can do some pretty amazing things. For example, fish can recognize human faces, use gestures to communicate, judge relative quantities (less/more) and form collaborations with other fish to achieve a common goal. How is all this possible with a small and simple brain? There are two options. 1) the tasks themselves are not as difficult as we think they are. It may be that the fish are using simpler solutions to solve what we consider a difficult problem. This would be interesting as it would mean we could find simple solutions to certain problems based on our knowledge of how a fish solves that problem. 2) Other structures in the brain are capable of performing the same functions of the neocortex. The neocortex is different from other brain structures because of the neurons and the connections between the neurons. If other brain sections could perform the same complex behaviours without neocortex neurons then it would make scientists rethink our understanding of the role of more complex brain structures.
Therefore, it is interesting to know both what fish can do as well as what they cannot and to understand how they are learning certain behaviours. I study what fish can learn using visual information. My research is divided into the following three themes:
Visually guided navigation
Being able to accurately and efficiently navigate from one location to another is necessary for the survival of many animals including some species of fish. I am interested in what information the fish use to navigate and what sorts of changes in their environment cause mistakes in navigation. More specifically, I am exploring how fish use landmarks as navigational beacons, how they decide what landmarks to use and if that landmark becomes unreliable, what other information can they use to navigate?
One way that a landmark can become unreliable is because of the movement of the fish itself. Recognition of 3D objects can be a difficult task as the appearance of some objects can change drastically depending on the angle from which they are viewed (see Object Recognition section for more information). Therefore, as a fish is swimming, the appearance of the landmark is constantly changing making it more difficult to recognize. Despite the difficulties, somehow fish are still able to navigate quickly and accurately. There are some behavioural adaptations that the fish could use which would make this task easier such as following the same routes over and over so they always see the landmark from the same orientation, or using landmarks that are farther away as these don’t change as much in appearance.
Because humans are constantly identifying objects around them based on visual information, one could be forgiven for assuming this is an easy task. However, object recognition is computationally difficult as eyes acts like a camera and only capture two-dimensional (2D) images of objects. Any change in lighting and orientation alter the 2D image, meaning the brain can face an almost infinite number of 2D images for a single object. Despite extensive research on the topic, how the brain is able to identify three-dimensional (3D) objects using 2D information is still unresolved.
Most research on the subject has come from studies with primates that have large brains. More recently, artificial intelligence systems have been applied to solve this problem but computers have not yet been able to fully replicate the accuracy of the animal recognition system. Fish have a proportionally small and simple brain compared to that of primates, yet are still capable of object recognition. Understanding how they do this may provide inspiration for models of object recognition that are simpler and less reliant on extensive processing capabilities.
Fish are being used to explore how animals make decisions that are vital to survival when the information upon which their decision is based is either incomplete or unreliable.