or trade-offs between the information received from different sensory systems are particularly important and fascinating. However, significant trade-offs also occur within a sensory system. In fact, compromises and trade-offs within a sense should be considered the norm. This reflects the simple truth that within a particular sensory system it is not possible to collect all of the information that is potentially available.
Trade-off within a sense is clearly seen in the relationship between visual resolution and sensitivity. At the very limit, both resolution and sensitivity are determined by the quantal nature of light and by noise within the nervous system. This trade-off is evident most dramatically in the fact that resolution always decreases as higher sensitivity is gained. We experience this every day of our lives: as light levels naturally fall, and our eyes become more sensitive to dim light, we accept that there is less detail in a scene. But the detail has not gone away, it is still there, it is just not detected by our eyes, which are adapted to detect the reducing number of light quanta in the environment.
Our loss of spatial details with decreasing light levels is not just a quirk of our vision. It is because it is a fundamental constraint on any vision system, including cameras. It is the very physical nature of light, its quantal nature, that precludes high visual resolution at low light levels. An eye that has evolved to achieve high sensitivity is unable to detect fine spatial information at low light levels, but neither can a highly sensitive eye readily achieve high resolution when there is a lot of ambient light. Life is full of compromises, and that is certainly true both within and between different sensory systems. Natural selection has worked on these trade-offs and fundamental constraints to shape sensory information optimally for the conduct of different tasks, in different environments, by different species (Figure 2.6).
FIGURE 2.6 The trade-off between high sensitivity and high resolution is found in all vision and imaging devices, including cameras and eyes. It arises from fundamental constraints imposed by the quantal nature of light. The trade-off is exemplified here by two bird species. In Short-toed Snake Eagles Circaetus gallicus vision has evolved to provide high resolution but low sensitivity, while in Tawny Owls Strix aluco high sensitivity is achieved but resolution is low. Hence, while the vision of a Tawny Owl is suitable for activity at low light levels it has low resolution and cannot detect fine details. On the other hand, eagles achieve high resolution but have low sensitivity and so they tend to cease activity as light levels fall towards dusk. (Photo of Short-toed Snake Eagle by A. Román Muñoz Gallego, University of Malaga.)
A unique property of vision
The trade-off between resolution and sensitivity exemplifies a unique and important aspect of vision. Within the range of light levels in which vision normally operates the information that it can extract varies markedly with the amount of the stimulus (light) that is available.
All other senses effectively provide the same range of information within their normal operating ranges. Across a wide range of sound levels, or concentrations of chemical compounds, or physical components of a touch stimulus, the sensory systems are able to extract more or less the same information. In vision this is not the case. This is because the information that vision provides is based primarily upon spatial resolution (the ability to see details in a scene or to say exactly where something is placed in the field of view), and resolution changes profoundly as light levels change (Box 2.3).
Box 2.3 Resolution and light levels
The effect of light levels on spatial resolution is significant. Acuity is often measured or estimated at high daytime light levels, but in a few bird species acuity has been measured across almost the full range of naturally occurring light levels (See Box 2.4 for a discussion of how natural light levels vary depending upon the elevation of the sun and moon). In some species acuity has been determined across the narrower range of the light levels that occur in daytime, and in some species across the daylight–twilight range.
This graph brings together data for a number of species and shows some key points about the effect of light levels on acuity. In all species shown here acuity has been determined using behavioural training techniques and gratings, the kind of technique depicted in Figure 2.7.
Although the maximum acuities of these different species are significantly different (for example, the maximum acuity of the eagle is 60 times higher than that of the dove) it is also clear that in all species acuity decreases considerably as natural light levels fall. The only species in which there is not a steep decline in acuity is the Western Barn Owl Tyto alba, but even in Barn Owls acuity shows a significant fall as light levels decrease to the lower ranges of night-time. In some species the decline is particularly steep, and this may be a significant reason why some daytime birds, such a doves, go to roost as light levels fall. If we wish to pose questions about how vision is used to control natural behaviours in any species, it is necessary to bear in mind that natural light levels have important effects on visual abilities, especially acuity.
The change in resolution with light level is significant. Unfortunately, this makes it very difficult for an observer to determine exactly what visual information could be available to a bird at any one moment. For example, we may have knowledge of what a bird can detect at one particular light level, but it will not be the same at another light level. Furthermore, colour vision is an important mechanism that enhances spatial resolution for a wide range of natural targets. However, colour vision functions only at high (daytime) light levels. As light levels drop to those of natural twilight and below, colour vision no longer functions and so an important source of information is lost. Furthermore, even if we discount the contribution of colour vision and just consider the ability to discriminate detail in black and white, spatial resolution decreases markedly with light levels.
At any one location naturally occurring light levels may change over a range of at least a million-fold (106) between maximum sunlight and moonlight. It is remarkable that our eyes can function across this whole natural range of light levels. Even more remarkable is that on moonless nights the range is extended downwards by a further 100-fold, and if we take into account how the presence of clouds and tree canopies can further reduce ambient light, then the total range of light levels in which an eye can function varies by a factor of 1011. That is a huge dynamic range for any detector (Box 2.4).
Box 2.4 Natural illumination sources and light levels
During the day (when the sun is above the horizon), under clear skies light levels vary by about 100-fold from sunrise to full overhead noonday sun. The coming and going of cloud cover can extend this range to 1000-fold. During the night, however, light levels can be much more variable. This is because the main source of light, sunlight reflected by the moon onto the earth, varies not just with the elevation of the moon, but also with the moon’s phase, which changes on a monthly cycle. This results in night-time light levels that can vary by 1 million-fold between sunset and starlight on a moonless night. On top of this there is a further potential 10-fold in variability brought by cloud cover.
This figure captures much of the huge range over which light levels naturally vary and why. It shows how natural levels of illumination at the earth’s surface depend upon the elevation of the sun and moon, and upon the phase of the moon. The amount of light from these sources changes continually over the daily cycle, and over a very large range. (Note that the scale of illumination is logarithmic,
