Feature

Reading a Spider's Mind

Good vision, flexible problem solving, deception and manipulation -- who can fathom the mind of a spider?

Robert R. Jackson and Duane P. Harland

Unique eyes, acute vision and complex behaviour are the distinctive features that separate the jumping spiders (of the family Salticidae) from all other spider families. Salticids enjoy a visual acuity approaching that of humans and exceeding that of any other animal of comparable size. One of our objectives has been to investigate the interrelationship between acute vision and complex behaviour in these unique animals.

We have been especially interested in Portia, a genus of salticids with exceptionally complex behaviour and perhaps the most acute eyesight of any salticid. The resolving power of Portias eyes is about two minutes of arc, or six times greater than the highest acuity known for insects (found in a large dragonfly with compound eyes roughly equal in size to Portias entire body) and is only six times less than that of humans. Although a spider is not a conventional subject of cognition studies, we have been investigating how Portias eyes work as part of a broader interest in animal cognition.

Most salticids eat insects captured in the open without using a web, but Portia is an oddball that routinely enters the webs of other spiders to catch and eat the resident. Hunting in the prey spider's own web is dangerous, but Portia avoids becoming its intended dinner's own dinner by using complex, flexible behaviour to deceive and manipulate its victim.

The web spiders on which Portia preys, having poor eyesight, perceive the world around them primarily by interpreting web signals. Web signals are the tension, movement and vibration patterns transmitted across the silk comprising the web -- the spider's web can be envisaged as not only a snare for catching prey but also a component of the web spider's sensory apparatus.

Portias success at araneophagy (or spider-eating) depends largely on being able to orchestrate the pattern of web signals received by the resident spider, a predatory tactic we call "aggressive mimicry". Using any combinations of its eight legs and two palps, Portia can produce a virtually unlimited array of web signals to control the behaviour of the resident spider prior to the attack [What is that Spider Thinking, Feb 95|].

Portias different prey spiders tend to be responsive to different signals, but Portia finds the appropriate signals by using a dynamic blend of pre-programmed tactics and trial-and-error derivation of signals. Trial and error is based on Portia using feedback from the prey spider to adjust the characteristics of the signals. Such flexible problem solving is perhaps surprising in a spider.

As another example of flexible problem solving, Portia routinely makes detours when it pursues prey. For instance, Portia may take a path to reach a particularly dangerous spider from behind. We know from experiments that many of Portias detours are planned ahead of time on the basis of preliminary viewing of the environment. Planned detours depend on acute vision. In fact, excellent eyesight is critical to much of the complex, almost mammal-like behaviour that makes Portia so fascinating.

Excellent Eyes

Salticids have eight eyes, but it is the large, forward-facing antero-medial eyes (principal eyes) that are responsible for acute vision. The other secondary eyes are primarily movement detectors. The principal eyes of salticids are very different from the multi-faceted compound eyes of insects. Instead, the salticid eye, like our own, has a single lens and a single retina. However, the way in which the salticid principal eye works differs from how a vertebrate eye works.

The salticid principal eye has a static corneal lens fixed to the carapace at the front of a long eye tube. In contrast, our own eyes are spherical with a corneal lens that moves with the rest of the eye and can be flexed during focussing. Unlike our own eyes, the salticid principal eye cannot accommodate -- it can not change focal length. Space is so limited in the salticid's small body, that lengthening the eye tube to focus is not feasible.

The principal eye is a telephoto system as a consequence of the eye tube being long and because there is a second lens at the back of the eye tube which magnifies the image from the corneal lens, turning this eye into a miniature telescope.

Within the salticid principal eye retina, photoreceptors are stacked in four layers at the rear of the eye tube, whereas our rod and cone photoreceptors are on one plane. The centre of the back-most layer is a fovea, a fine-grain, regular mosaic of receptor cells where small inter-receptor angles maximise acuity. This fovea contains only a few hundred receptors, and the field of view covered by the fovea is only a small part (about 2^o) of the field covered by the corneal lenses of the principal eyes (about 25^o). Six muscles attached to the outside of the eye tube allow the salticid to sweep the fovea's field of view over the scene coming through the fixed corneal lens.

Our knowledge of salticid eyes comes especially from the ground-breaking research by Michael Land of Sussex University, England, carried out 30 years ago, and more recent work by David Blest at Australian National University. Land suggested that the intricacy of the eye-tube movement may be part of the mechanism by which the salticid perceives shape and form, but precisely how this might be achieved is poorly understood.

One exciting possibility is that the salticid, by adopting particular patterns of eye-tube movement, may be searching for specific identifying features of the object being viewed. The behaviour of the salticid's eyes may reveal how perception is achieved. Yet 30 years later, Land's pioneering study is still almost everything we know about salticid eye-tube movement. Methodological difficulties, including the need for a specialised opthalmoscope, have probably discouraged further research.

Recently, we devised a system for studying how eye-tube movement may function in the processing of optical cues, and one of us has built a prototype opthalmoscope based on Land's design. Our goal is to record pattern of eye-tube movement while Portia views the objects we place in its field of view during experiments.

The apparatus used to study salticid eye movement. The spider is given an object to look at. Red light, which is invisible to the salticid, is used to illuminate the eye tubes and observe their movement.

Our goal is to go beyond studying reception, a term for when an animal takes in raw information from the sense organs. We are also interested in representation, a term for a cognitive level one step beyond reception. This term refers to the moulding of raw sensory input into what is needed for identifying objects and solving problems.

The conventional wisdom used to be that representation in salticids is based on the use of only a few simple optical cues to discriminate between objects belonging to only a few broad categories. In reality, the model implied is probably far too simplistic for any salticid, and representation is certainly much more complex than this in Portia. This is illustrated by examples from recent work on the things Portia can distinguish:

• insects from spiders, regardless of whether the two prey are in or out of webs
• flies (on which Portia preys) and ants (which Portia avoids)
• different species of spiders
• the spider and the spider's eggs
• spiders that are feeding on their own prey (insects) and spiders that are not feeding
• the orientation of the spider (whether it is facing forward or away)

It is Portias large repertoire of distinct behavioural responses that enables us to ascertain when discriminations are made, because in each instance performance of a different behavioural sequence provides the objective evidence that Portia has made a discrimination. Movement of the object is unnecessary for any of these discriminations, chemical cues are ruled out by the experimental design and, in general, shape and form alone appear to be sufficient. A more complex model has been suggested, but even this is almost surely too simplistic.

We are currently investigating the optical cues by which Portia makes these and other discriminations. Experimental protocol includes presenting Portia with models made from dead spiders and insects that we mount in lifelike postures on small pieces of cork. Features of the models are altered systematically. For example, we alter the size of the eyes and the length and orientation of the legs. Also, one of us has also developed a system for testing Portia with virtual objects generated by computer 3-D animation and displayed to Portia through a high resolution projector. Using this system, which is a first of its kind for studies of salticid vision, we can achieve very precise control over the optical cues given to Portia.

Some preliminary findings have been intriguing.

Most spiders have eight eyes, but salticids are unique because their two antero-medial eyes are much larger than the other six. This is an important taxonomic character for distinguishing salticids form other spiders. Our findings indicate that Portia, like the human taxonomist, relies on the relative size of the spider's antero-medial eyes when distinguishing salticids from other spiders.

Portia is itself a salticid, yet Portia responds differently depending on whether the salticid it encounters is or is not another Portia. Important cues include distinctive tufts of hair on Portias legs which are absent from the legs of more typical salticids.

Pholcids, web-building spiders with especially long legs, are common prey of Portia. When Portia contacts a pholcid's leg, it often gets wrapped up and eaten. Portia compensates by being especially careful to achieve an orientation from which the pholcid's body can be attacked without contacting a leg. Important cues for recognising a pholcid include presence of legs that are at least five times longer than the body.

About 50 years ago, Keith McKeown, an Australian naturalist, asked rhetorically, "Who can fathom the mind of a spider?" When we first began to study spider behaviour, McKeown's question struck us as almost comical. Our attitude has changed over the years, and we are now taking seriously questions about spider cognition. Perhaps, no one will ever fully fathom the mind of a spider, but the question no longer appears so foolish as we might first have thought.

Mirroring a Spider's Mind

Duane Harland works in the Department of Zoology at the University of Canterbury
Robert Jackson is senior lecturer in zoology at the University of Canterbury.