Goalkeepers doubtless need excellent responses to prevent the opposition from scoring that goal. However, exceptional responses are not enough. To reach and catch the ball at the right moment, the goalkeeper has to predict where the ball will end up and calculate where their movements will take them to capture the ball. This type of control has been found only in vertebrates so far. However, a recent study published in Nature by a team of researchers led by Anthony Leonardo from the Howard Hughes Medical Institute, USA, has demonstrated that dragonflies that are in pursuit rely on internal models that predict the effects of movements of their own body and their target, like the goalkeeper catching the ball.
Dragonflies are brilliant aeronauts that can hover, fly at high speed and perform agile manoeuvres to defend their territory, mate on the wing or chase prey. Their vision is also excellent: with their large eyes they can see in almost any direction. Usually they lurk on plants where they wait for prey insects to fly over. Once the prey is in focus, the dragonfly rapidly lifts off and manoeuvres to approach the prey at an angle from below, so that it can trap the victim with its capture legs.
To analyse this type of hunt in more detail, the team filmed dragonflies in slow motion during pursuits. It became obvious that the dragonflies did not rely exclusively on their reactions, because they did not always respond to unexpected changes in the prey motion. Rather, they tried to align their elongated body with the flight path of their prey to approach and strike from below, minimizing the chances of being discovered. To do so, the head of the dragonfly moves independently from the body, so that the eyes can continuously focus on the prey while the body is aligned with the prey's flight path.
The scientists then dissected the head movements in more detail. They set up a flight arena, which allowed them to record the paths of dragonflies and prey with high accuracy using high-speed cameras. They also placed micro-reflective markers on the head and body of the dragonflies to record the relative movements. Based on these data, they then calculated the angular position of the prey image on the dragonfly's eye.
What they found was quite surprising, as the head motions turned out to compensate precisely for drift of the prey image on the eye that resulted from the dragonfly's own body movements and the anticipated motion of the prey. The high synchrony and precision of the timing of these head movements suggest that dragonflies use internal calculations to generate models that predict how body and prey movements will influence the position of the image on the dragonfly's eye, and how the head must then be moved to cancel out these effects: classical sensory feedback. This predictive system largely compensates for the dragonfly's own body movements and thus relieves the visual system to detect sudden prey manoeuvres, to which the dragonfly can respond by reactive control.
Leonardo and his team have shown for the first time that invertebrates use internal models to predict the effects of their own body movements when targeting prey. The fact that all of the experiments were done under laboratory conditions where the prey's movements are more restricted may suggest that this predictive steering control could be dominated by reactive mechanisms in the wild, thus explaining why it had been overlooked for so long.
