Box 3.4 Advantages of behavioural hierarchies
3.2 Sensory mechanisms, perception and behaviour
3.2.1 Sensory mechanisms, stimulus filtering and perception
For an animal to behave appropriately, its nervous system must receive the right kind of information. It must be able to register relevant changes in the environment. Several types of change may be important to the animal’s survival and reproductive success.
Changes in light intensity, temperature, sound, tactile stimuli, odours, barometric pressure or other factors may signal the approach of a predator, a potential mate or prey item, or a critical change in weather conditions. All therefore need careful mon-itoring. How have the various sensory mechanisms evolved to cope with this monitoring task?
3.2.1.1 Visual stimulus filtering
At one time, the eye was thought of as little more than a means of translating images of the environment into electrical impulses. It was only in the brain that information was sorted and interpreted. A ground-breaking study of retinal function in the leopard frog (Rana pipiens) by Lettvin et al. (1959), however, changed all that. Earlier studies of the leopard frog had looked at its responses to points of light and dark, and concluded that the frog’s eye simply registered changes in the tone of the viewed object. Lettvin et al., however, stimulated the retina with images possessing some of the basic features of natural objects that were salient to the frog in its day-to-day life. By this means they discovered a remarkable apparent functional diversity among the ganglion cells of the retina (Table 3.1). The qualification ‘apparent’ is important because later work showed that, while retinal ganglion cells undoubtedly send important perceptual information to the brain, they do not quite fulfil the role of specific stimulus detectors that Lettvin et al.’s classification might suggest (Ewert 1997).
Studies of a variety of frog and toad species have shown that visual information from the retina is sorted by different classes of ganglion cell according to their different sensitivities to the size, contrast, motion, colour and edge characteristics of stimuli. This 3.2 n Sensory mechanisms, perception and behaviour x 133 AB_C03.qxd 9/17/07 8:05 PM Page 133
structured information is received by neurons in the optic tectum and pretectal thalamus and coordinated to generate appropriate motor output. Jörg-Peter Ewert’s studies of the common toad, Bufo bufo, show how the relationship between peripheral and central processing works.
In a series of classic experiments, Ewert (1974, 1980) probed the events in the eye and brain of the toad that determined its response to different kinds of visual stimuli.
Using microelectrodes to record from single cells in the tectum of freely moving toads, he showed that visual perception depended on a combination of peripheral and central stimulus filtering. Peripheral filtering occurs via the structured receptive fields of the retinal ganglion cells (whose axons extend down the optic nerve to the brain), each of which receives input from a small elliptical area of the retina. Stimulation of the ellipse sends impulses to the appropriate ganglion cell via bipolar cells which determine whether or not the ganglion cell is likely to fire. However, the receptive field of each ganglion cell is divided into two regions: a central excitatory region, which is stimulated to respond when it receives input from bipolar cells, and a peripheral inhibitory region, which reduces the likelihood of the ganglion cell firing when stimulated. The extent to which objects passing across the toad’s visual field stimulate the central and peri-pheral receptive fields therefore determines whether the cells pass information back to the brain. A small beetle moving across the toad’s field of view is likely to stimulate the central fields of the ganglion cells but leave the peripheral fields unmoved, thus eliciting a stream of messages back to the tectum. A large object looming close to its eye, on the other hand, will stimulate both central and peripheral fields, thereby inhibiting a response. In effect, a toad sees only those things that change the light intensity falling on its ganglion cells: in the main, small moving objects, such as beetles, flies or distant pre-dators, that impact on its chances of survival. Large, stationary images are more likely to be rocks or tree stumps, or some other inconsequential object unworthy of a response.
While peripheral filtering goes on in the retina, neurons within the optic tectum receive information from clusters of neighbouring ganglion cells. Thus, each tectal neuron has its own receptive field based on the area of the retina serving its ganglion cells. Pathways also extend from the retina to cells in the thalamus, with additional connections between the thalamus and the tectum. Ewert investigated the receptive fields of individual cells in the tectum and thalamus by implanting an electrode and recording the responses of the cells when different objects were passed in front of the toad’s eyes. The results Table 3.1 Lettvin et al.’s (1959) functional classification of retinal ganglion cells in the leopard frog (Rana pipiens). See text
1. Sustained-edge detectors showed the greatest response when a small, moving edge entered and remained in their receptive field. Immobile or long edges did not evoke a response.
2. Convex-edge detectors were stimulated mainly by small, dark objects with a convex outline.
3. Moving-edge detectors were most responsive to edges moving in and out of their receptive field.
4. Dimming detectors responded most to decreases in light intensity.
5. Light-intensity detectors: the responsiveness of these cells was inversely proportional to light intensity. They were most responsive in dim light.
showed a variety of responses to the different objects, some very strong, others weak or non-existent. These variations create a further, central, tier of stimulus filtering, with different cells responding more strongly to objects of different shape or orientation.
Some cells in the tectum, for instance, are most responsive to long, thin objects moving horizontally across the toad’s field of view (Fig. 3.21a). Others in the thalamus respond most to objects moving through in a vertical orientation (Fig. 3.21b). Electrical stimulation of tectal and thalamic cells has shown that excitation in the tectum results in the toad orientating and leaning towards the perceived object, opening its mouth and snapping the object up with its tongue, and performing actions associated with cleaning its mouth (Ewert 1974; Fig. 3.22a). Excitation in the thalamus, on the other hand, elicits defensive crouching, rising up or avoidance behaviours (Fig. 3.22b). Ewert characterises these opposing sets of responses in terms of potential ‘prey’ and ‘enemy’ (respectively) stimuli.
However, excitatory and inhibitory connections between the thalamus and tectum are also important in modulating the toad’s response.
3.2 n Sensory mechanisms, perception and behaviour x 135
Figure 3.21 Visual stimulus filtering in cells of the tectum and thalamus of the toad Bufo bufo. Electrical recordings show that cells in the tectum are stimulated most strongly by a moving object extended in the direction of movement (a). Cells in the thalamus, however, respond most to objects extended perpendicularly to the direction of movement (b). From Ewert (1974).
Figure 3.22 (a) Electrical stimulation of the optic tectum (a visual centre in the brain) in toads elicits a sequence of prey-catching behaviours such as snapping, while (b) stimulation of the thala-mus elicits defensive movements. After Ewert (1974).
AB_C03.qxd 9/17/07 8:05 PM Page 135
On the basis of information from retinal ganglion cells, therefore, the optic tectum tells the toad where in the visual field an object is located, how large it is, how fast it is moving and how much it contrasts with the background, while connections between the tectum and thalamus refine perception to allow the toad to assess the significance of the visual signals. The filtering process can thus be envisaged as a discriminatory cascade, each tier of which analyses and amplifies a particular aspect of the object in view.
In mammals, central processing of visual stimuli follows similar principles. The visual cortex of cats, for example, contains two main types of cell. ‘Simple’ cells respond to lines and edges in particular orientations or locations via excitatory ‘on’ and inhibitory
‘off’ zones within their receptive fields, somewhat akin to the retinal ganglion cells in frogs and toads. ‘Complex’ and ‘hypercomplex’ cells also respond to lines, slits and edges in different orientations, but are not divided into ‘on’ and ‘off’ zones. Instead, the whole unit increases or decreases its rate of firing depending on the kind of input.
‘Complex’ cells receive information from several ‘simple’ cells, and filtered responses are passed on from the visual cortex to other parts of the brain. The recognition of complex visual stimuli thus depends on neural activity at many stages along the visual pathway.
Although cats rely more on central rather than a peripheral filtering process, there is not a simple evolutionary progression towards centralisation from lower to higher organisms. Pigeons, for example, show even more retinal differentiation than the leopard frog, while many invertebrates, such as crabs (Podophthalamus vigil) and locusts (Locusta migratoria), have sophisticated central processing. The selective elicitation of behaviour as a result of central stimulus filtering in part prompted the concept of innate releasing mechanisms (IRMs) in ethology (see 3.1.3.2).
Perceptual rules of thumb
Visual stimulus filtering provides common toads with a rough but workable guide to what is edible and what should best be avoided. We can thus think of the toad having perceptual rules of thumb (see 2.4.4.3). Like all rules of thumb, the toad’s ‘prey’ and
‘enemy’ rules work well enough in the world in which the animal normally operates, but can easily be fooled by novel cue configurations or experimental manipulation. Thus the toad’s ‘enemy’ response to a snake with a raised head (Fig. 3.23a) can be elicited by an abstract pattern with a raised element (Fig. 3.23b) or even a leech, which is normally regarded as prey, if its front suckers are raised off the ground (Figs 3.23c,d) (Ewert
& Traud 1979).
Our own visual perception is also heavily dependent on rules of thumb. Our pro-pensity for visual illusions tells us a great deal about the filtering processes and rules of interpretation that determine the model our brains build of the world. As Richard Dawkins (1998) puts it, the brain is a natural onboard virtual reality computer, con-structing images of the world according to rules honed by natural selection. Information is amplified, integrated, suppressed or synthesised to generate the best working hypo-thesis for functioning in the environment. Thus, a glimpse of fur in the undergrowth is extrapolated into a predator, images of different size translate into perspective, subtle changes in facial expression become beacons of social information. Several excellent illustrations can be found in Gregory (1998).
The brain as hypothesis generator is exemplified particularly clearly when it dithers between equivalent alternatives. A good example is the Necker cube (Fig. 3.24a), a simple two-dimensional set of lines of paper which the brain interprets as a three-dimensional
cube. Stare at the apparent cube for a few seconds, however, and it flips in depth per-spective between two alternative forms, each interpreting a different ‘end’ facet as being to the fore. Being faced with two equally valid alternatives, the brain flips backwards and forwards between them rather than plumping arbitrarily for one. Familiar ambiguous images illustrate the same principle (Fig. 3.24b).
The capacity for interpretation is illustrated by Fig. 3.24(c), in which we perceive a white shape lying across the triangles where in fact none exists. The broken lines of the triangles, and the white segments of the circles, are suggestive enough for the brain to invent the rest. Our ability to recognise faces or familiar scenes in simple line sketches relies on the same inventive filling-in. But why does the brain do this? Because we inhabit a world of objects, and objects have boundaries that distinguish one from another.
Anything that suggests the presence of an object is worth noting, and if necessary extrapolating, because many objects have salience: they are food, predators, companions or dangerous obstacles. We overlook them at our peril.
The adaptive value of these perceptual tricks is that they allow the brain to rationalise the environment according to well-tried rules of operation, often on the basis of partial or ephemeral information and with limited time to waste. Like all rules of thumb (2.4.4.3), they are prone to error, but as long as errors are sufficiently rare, and/or are not too costly, they provide an economical but effective means of interpreting the world.
3.2 n Sensory mechanisms, perception and behaviour x 137 Figure 3.23 Visual rules of thumb in the
toad Bufo bufo. ‘Enemy’ responses are pro-voked by images of a snake (a), a head-rump dummy (b) and a leech with a raised front sucker (c). If the leech’s sucker lies in the plane of movement (d), however, the toad responds as if it is prey. From Ewert (1980).
AB_C03.qxd 9/17/07 8:05 PM Page 137
3.2.1.2 Auditory stimulus filtering
Like vision, hearing has become specialised in different species according to the demands of their way of life, a process that also frequently relies on stimulus filtering.
In some cases, filtering is achieved by adjusting the mechanics of the hearing apparatus to focus on specific components of the sound environment. This is particularly clear in nocturnal hunters such as bats and owls, some of which show extraordinary sensitivity to the sounds generated by their prey. Pioneering work by Payne (1971), for example, has shown that barn owls (Tyto alba) can home in on the faint noises made by mice as they move through the litter or gnaw their food. Even barely audible (to humans) sounds such as a leaf being pulled across a floor can be pinpointed. Barn owls achieve this remarkable accuracy by means of the positional asymmetry of the ears on either side of the head (the left ear is higher than the mid-point of the eye, the right lower) and the arrangement of feathers around the face, which form the facial ‘disc’ (Fig. 3.25a). The facial disc helps to channel incoming sound into the auditory meatus of the two ears and filter out sounds that are not arriving along the line of vision. Information about the prey’s position in the horizontal plane can then be gleaned from the relative stimulation of each ear (Fig. 3.25b), while the asymmetric positioning of the ears provides informa-tion in the vertical plane (Knudsen & Konishi 1979).
Auditory stimulus filtering in barn owls involves mainly the removal of directional redundancy. Among insects, however, there are many examples of ‘ears’ which respond solely to a limited range of sound frequencies. One of the best known comes from noctuid moths. These moths are heavily preyed upon by night-flying bats that use echolocation to identify and home in on their airborne prey. Echolocation in bats works like human sonar in that it relies on the animal emitting pulses of sound and listening to their echoes as they bounce back from the environment. The pattern of returning echoes enables the bats to create a ‘sound topography’ of the environment by which they can orientate and navigate their way around. Objects distort the returning sound waves in different ways, allowing bats to judge their size, shape and texture as well as, in the case of moving objects, their speed and direction of movement (Simmons & Stein 1980). Whether or not an object will generate an echo, however, depends on its size and the wavelength of the sound. The wavelength must be roughly equal to the diameter of the object to produce an Figure 3.24 Visual illusions. (a) A Necker cube: a two-dimensional apparent cube that appears to flip between downward left and upward right orientations. (b) A classic ambiguous image alternat-ing between a chalice and two opposalternat-ing faces in profile. From Eye and Brain: The Psychology of Seeing by R.L. Gregory (Fifth Edition, 1998). Reprinted by permission of Oxford University Press.
(c) An illusion of a second triangle overlying the one outlined. From Marr (1982).
echo. Since the aerial prey of most bats are small insects, bats are constrained to use very high frequency (ultrasonic) sound, somewhere in the region of 50 – 80 kHz in fact, well above the upper limit (around 20 kHz) of human hearing. While humans may not be able to hear bats, however, some of their insect prey certainly can, noctuid moths among them. In fact, the ears of noctuids are almost entirely geared to listening out for bats.
The moths possess a pair of ears, one either side of the thorax. Each ear consists of just two sensory neurons connected to a tympanum, one sensitive to low-intensity sound, the other to high-intensity sound (Roeder 1970). The low-intensity (A1) neuron also responds more to intermittent pulses of sound rather than continuous bursts. Neither low- nor high-intensity (A2) neurons, however, respond to different frequencies of sound, only their intensity and temporal pattern of emission. The A1 neuron responds to the faint ultrasonic emissions of bats up to 10 m away, and fires as the emissions increase in intensity. The moth can therefore tell whether the bat is getting closer.
By comparing the relative stimulation of the A1 neuron in its two ears, the moth can also pinpoint the position (above, below, to the side) of the bat and move away. The A2 neuron comes into play only when the bat flies close to the moth and the neuron is exposed to very high-intensity emissions. When the A2 neuron fires, impulses are trans-mitted to the cerebral ganglia and inhibit the centre controlling activity in the thoracic ganglion. The thoracic ganglion controls the pattern of wing beat, so, when it is inhibited, the wings beat asynchronously or stop beating altogether. As a result, the moth either flies erratically or drops like a stone out of the flight path of the bat.
Noctuid moths therefore show an extreme degree of auditory stimulus filtering.
In the myriad noises of the night, their hearing system is tuned exclusively to the sound signatures of their predators. While motor responses are mediated by neurons in the 3.2 n Sensory mechanisms, perception and behaviour x 139
Figure 3.25 (a) The facial ‘disc’ of the barn owl (Tyto alba) is a fan arrangement of feathers around the face that helps to channel sound to the asymmetrically positioned ears. (b) Barn owls match the sound input to each ear to locate and catch prey in the dark. A mouse at A will stimulate ear 1, but not ear 2, and vice versa at C. A mouse at B, however, will stimulate both ears. From Barnard (1983) after Alcock (1975).
AB_C03.qxd 9/17/07 8:05 PM Page 139
CNS, filtering is achieved peripherally via the highly selective neurons of the ear. An intriguing example of auditory peripheral filtering in vertebrates comes from the tree frog Eleutherodactylus coqui, whose specific name derives onomatopoeically from the characteristic ‘co-qui’ call of the male. The call is particularly interesting because its two components (‘co’ and ‘qui’) appear to be directed at different recipients – males and females respectively. Selective targeting could be achieved in a number of ways.
However, Navins & Capranica (1976) discovered that the tympanic membranes of the two sexes are in fact tuned differently, so that each sex hears only the relevant part of the call. Males hear the ‘co’, females the ‘qui’, and neither hears the other component.
In mammals, centrally mediated selective attention appears to be the more important mechanism for biasing responses to auditory stimuli.
3.2.1.3 Other mechanoperception
Hearing works through high-frequency stimulation of mechanoreceptors, structures (such as the tympanum of the insect and mammalian ear in the case of hearing) that are sensitive to vibrations caused by pressure changes. Many kinds of mechanical stimuli are important to animals and they have evolved a wide range of receptors and organs to
Hearing works through high-frequency stimulation of mechanoreceptors, structures (such as the tympanum of the insect and mammalian ear in the case of hearing) that are sensitive to vibrations caused by pressure changes. Many kinds of mechanical stimuli are important to animals and they have evolved a wide range of receptors and organs to