174
Simon B. Laughlin
components (molecules, cells, and cell assemblies), combine with the specificity
and spatial precision of developmental mechanisms. Together they enable
organisms to generate subtle and ingenious solutions to the practical problems of
collecting physical and chemical energy, transducing this energy, and gaining
information from it.
Engineering for high performance is illustrated by two examples from optics.
The nocturnal spider Dinopis depends upon night vision to locate its prey.
Dinopis maximizes the sensitivity of its photoreceptors by using an exceptionally
powerful lens, with F-ratio 0.58, to concentrate the image on its retina. The severe
spherical aberrations of the lens are reduced by constructing a refractive index
gradient within its bulk (Blest and Land 1977). Similar gradients reduce
aberrations in mammalian lenses (Hughes 1977). The second example is the deep
pit found in the retinas of raptors, chameleons and jumping spiders. Each of these
animals requires high spatial resolution to locate prey. To maximize the
anatomical resolving power of photoreceptors, the eye is made as large as
possible and spans the relevant body cavity. The pit has a similar profile in each
animal because it is constructed to form a diverging lens (Williams and Mcintyre
1980). This lens increases spatial resolution beyond the limit set by eye length, by
converting the eye into a telephoto system.
Faced with numerous examples of this type (Walls 1942; Hughes 1977; Land
1981; Martin 1983; Nilsson 1989), we must agree with Charles Darwin that
many eyes are "organs of extreme perfection and complication" (Darwin 1859).
Perfection points to the selective advantage of good vision, and suggests that
natural selection is favouring organisms that operate efficiently within physical,
biophysical and developmental constraints, to make the best use of their
resources. Perfection is not limited to optics, it extends to neural coding.
5 The Optimization of Neural Coding
Neural coding in the retina provides a compelling example of the efficient use of
a limited resource. A given retina contains a certain number of photoreceptors
and neurons, each with a limited ability to transmit information. A cell's
information capacity is ultimately limited by the number of different messages
that it can transmit in a given time, as determined by two factors, the number of
discriminable signal levels, and the rate at which the cell can change these levels.
In electrical signalling both of these factors are constrained by elementary
biophysics (Laughlin 1989). The maximum number of discriminable levels is the
cell's response range divided by the amplitude of the noise. Noise is always
present because it is generated by the elementary molecular events that produce
and process cell signals (e.g. photon absorption, synaptic vesicle release, binding
and dissociation of transmitter, channel opening). The rate at which receptors and
neurons can change signal level is limited by the membrane time constant.
Simon B. Laughlin
components (molecules, cells, and cell assemblies), combine with the specificity
and spatial precision of developmental mechanisms. Together they enable
organisms to generate subtle and ingenious solutions to the practical problems of
collecting physical and chemical energy, transducing this energy, and gaining
information from it.
Engineering for high performance is illustrated by two examples from optics.
The nocturnal spider Dinopis depends upon night vision to locate its prey.
Dinopis maximizes the sensitivity of its photoreceptors by using an exceptionally
powerful lens, with F-ratio 0.58, to concentrate the image on its retina. The severe
spherical aberrations of the lens are reduced by constructing a refractive index
gradient within its bulk (Blest and Land 1977). Similar gradients reduce
aberrations in mammalian lenses (Hughes 1977). The second example is the deep
pit found in the retinas of raptors, chameleons and jumping spiders. Each of these
animals requires high spatial resolution to locate prey. To maximize the
anatomical resolving power of photoreceptors, the eye is made as large as
possible and spans the relevant body cavity. The pit has a similar profile in each
animal because it is constructed to form a diverging lens (Williams and Mcintyre
1980). This lens increases spatial resolution beyond the limit set by eye length, by
converting the eye into a telephoto system.
Faced with numerous examples of this type (Walls 1942; Hughes 1977; Land
1981; Martin 1983; Nilsson 1989), we must agree with Charles Darwin that
many eyes are "organs of extreme perfection and complication" (Darwin 1859).
Perfection points to the selective advantage of good vision, and suggests that
natural selection is favouring organisms that operate efficiently within physical,
biophysical and developmental constraints, to make the best use of their
resources. Perfection is not limited to optics, it extends to neural coding.
5 The Optimization of Neural Coding
Neural coding in the retina provides a compelling example of the efficient use of
a limited resource. A given retina contains a certain number of photoreceptors
and neurons, each with a limited ability to transmit information. A cell's
information capacity is ultimately limited by the number of different messages
that it can transmit in a given time, as determined by two factors, the number of
discriminable signal levels, and the rate at which the cell can change these levels.
In electrical signalling both of these factors are constrained by elementary
biophysics (Laughlin 1989). The maximum number of discriminable levels is the
cell's response range divided by the amplitude of the noise. Noise is always
present because it is generated by the elementary molecular events that produce
and process cell signals (e.g. photon absorption, synaptic vesicle release, binding
and dissociation of transmitter, channel opening). The rate at which receptors and
neurons can change signal level is limited by the membrane time constant.