"It has long been obvious that sensory modalities go well beyond the classic five human senses of hearing, sight, taste, smell, and touch. Living things not only sense sound, light, chemicals, and pressure, but also position, heat, gravity, acceleration, electrical and magnetic fields, and even the passage of time. A glance at a list of some of the better-studied of these sensory systems (Table 1) may lead to the impression that life has evolved to monitor just about everything. Can there be any unifying themes, any biophysical principles?

A physicist looking over Table I might point out that living things sense manifestations of just two of the four fundamental forces: the electromagnetic force and the gravitational force. The electromagnetic force has infinite range, and it dominates on the length scale at which life exists: from nanometers up to tens of meters. It holds molecules together, and is the basis for light, heat, sound, and all of chemistry. The gravitational force also has infinite range, and, although its effects are far weaker, it, too, affects living forms. But the remaining two forces, the strong force (or nuclear force) and the weak force (responsible, for example, for ß-decay), appear to pass undetected through the biosphere. It is likely that, because the range of both the weak and strong forces is finite and short (small, even on the scale of an atom), large-scale consequences are wholly insignificant. Perhaps so. But who knows? Someone may yet find an example of a creature that detects nuclear processes or violates parity conservation."

"Table 1
A List of Some of the Better-Studied Sensory Modalities, in No Particular Order

vertebrate rod and cone vision        vertebrate taste transduction
chemotaxis in bacteria                   animal map senses
vertebrate hearing                        magnetotaxis in bacteria
echolocation in bats, birds              magnetoreception in invertebrates
taste reception                            electroreception in fish
chemotaxis by eukaryotic cells        ultraviolet light detection
tactile responses of protozoans       insect chemical signaling
vertebrate olfaction                      pH taxis in microorganisms
odorant detection insects              insect pheromone detection
yeast mating response                  sense of elapsed time
circadian rhythms                         insect rhabdomeric vision
osmotic responses of bacteria        stretch-inactivated receptors
salt taxis in bacteria                     geotaxis in microorganisms
phototropism in plants                   phototaxis in protozoa
phototaxis in bacteria                   thermoreception in vertebrates
stretch-activated receptors           aquatic bouyancy regulation
vestibular senses                          infrared sensing and imaging
insect tactile/vibration responses    polarized light detection
geotropism in plants                      infrasound detection
proprioreception                           leukocyte chemotaxis/signaling
magnetoreception in vertebrates     sonar in marine mammals
gas partial pressure sensing            fluid or gas velocity detection
haltere-based orientation               fungal avoidance response
thermotaxis in microorganisms         osmoregulation in plants
crustacean rhabdomeric vision        nociception    "

"It turns out that the question of "optimality" is ill posed. There are a number of reasons for this. First, and almost trivially, optimality supposes that a unique solution exists that maximizes the performance of a sensory system. In fact, there may well be multiple solutions to a sensory problem, any one of which achieves the desired level of perfection. The incredible natural variety of sensory systems reminds us that there are many ways to skin a cat. Second, there is no a priori reason to believe that optimality has been achieved. On the contrary, to do so would be tantamount to assuming that evolution had somehow run its course and produced a final product. It is arguably better to think of biological systems as "works in progress." Third, and of fundamental importance, one cannot talk about optimization without first stipulating (1) the properties (functions) that are to be optimized, and (2) all the constraints (boundary conditions) for that optimization. This is where one rapidly gets into trouble with biological systems. It is simply not meaningful to say that performance is maximal unless a context is specified. Just what measure of performance is appropriate (signal amplitude? signal-to-noise ratio? speed? jitter? encoding fidelity?) and what factors contribute to the "design criteria" (basic physics? environmental factors? size? metabolic cost? selective advantage?)? As researchers on the outside looking in, we should view our task as being the identification of precisely these things: only then can questions of optimality be addressed meaningfully. To put it all in a more "biological" language:

Evolution doesn't really seek to optimize. It seeks to iterate, to ramify, and to compromise.
The solutions found by evolution are neither unique nor perfect.

A corollary of this, therefore, is that:

Sensory systems are not necessarily as good as they can be. They are just as good as they
need to be."

"
The Importance of kT
The thermal energy, E_therm , associated with an absolute temperature, T, is given by kT, where k is Boltzmann's constant. Boltzmann's constant equals the universal gas constant, R, per molecule, i.e., k = R/N_A, where N_A is Avogadro's number: it has the value k = 1.38 x J/° K^-l, or 1.38 x 10^-16 ergI/° K^-l. Room temperature, -25°C (- 300K), corresponds to a thermal energy E_therm= 4 x 10^-21 J = 4 x 10^-14 erg.
Physicists tend to think of this energy as - 1/40th of an electron volt (0.025 eV). In chemists' units, that comes to -0.58 kcal/mol. Classically, by the equipartition theorem, all bodies in thermal equilibrium have 1/2 kT of energy per degree of freedom. A biological sensor carries at least this much energy as a baseline level. Additional energy deposited by a sensory signal therefore falls on a system that is already energized by thermal noise. Whether or not the signal can be detected will depend on how much energy it carries compared with the thermal background, as well as how much time the detector has to make the measurement. For a discussion of these considerations, the reader is encouraged to consult the excellent review by Bialek (1987), from which portions of the following discussion were drawn.

"
Electroreception Limits
Many aquatic organisms respond to electric fields, especially electric fish, sharks, skates, and rays (Kalmijn, 1982; Heiligenberg, 1984; Bullock and Heiligenberg, 1986). This is hardly surprising, since oceans are filled with electrical signals containing useful information. Freshwater fish live immersed in a weakly conducting medium, and saltwater fish live in a rather better conductor. The weakly electric fish navigate, communicate, and locate using electrical signals. Electrical responses have also been observed in insects, pigeons, and other terrestrial organisms. Electrostatic fields near the earth's surface generate field gradients ~1 V/cm in air, while electrical storms can produce fields in excess of 10 V/cm. Even isolated chicken or bovine fibroblasts have been found to react to weak oscillatory electrical fields. What sets the limits for electroreception?

Most cells respond to oscillating (AC) electrical fields in the range 10^-2-10^-4 V/cm. But sharks, in particular, respond to fields as low as 10^-9 V/cm (Kalmijn, 1982)! The speeds of electrical oscillations vary widely: electric fish produce sharply pulsed fields with repetition rates from 5 to 3,000 Hz, with 200-400 Hz being typical. This falls into roughly the same range of frequencies as human hearing. Electrical impulses generated by the firing of nerves in swimming creatures also tend to fall in the audio range. So, following the same logic as with hearing, the energy of the quantum associated with these fields, hv, will be about ten orders of magnitude less than thermal energy kT: electroreception is dominated by thermal, not quantum, effects."

"
Magnetoreception Limits
Sharks, skates, and rays are so sensitive to electrical fields that they are able to sense the earth's magnetic field through electromagnetic induction generated by their movements across magnetic flux lines. Their magnetoreceptive mechanism is indirect. On the other hand, birds, bees, butterflies, salmon, tuna fish (and probably a host of other organisms) are able to detect magnetic fields directly, probably by means of ferromagnetic or superparamagnetic detectors. The cellular basis of this form of magnetoreception remains a mystery, and magnetosensory organs (or cells) have yet to be identified. But here are some data worth pondering. The earth's magnetic field is about 0.5 Gauss at the surface (a note on units: 0.5 G = 50,000 γ = 50 pT; hence γ's equal nanotesla, nT). Its strength increases by some 3-5 γ per kilometer from the equator to the geomagnetic pole. It varies periodically, the circadian variation being 10-100 γ (it also reverses polarity chaotically every 10,000- 100,000 years or so). Magnetic storms produce fluctuations of 10-3,000 γ. Terrestrial anomalies (iron deposits and such) represent deviations of 30-30,000 γ. To employ magnetic fields for serious navigation (i.e., determination of fractions of a kilometer), as pigeons and bees apparently do, a sensitivity of several γ would seem to be in order. In fact, behavioral thresholds have been measured at 5-20 γ for pigeons and 1-10 γ for honeybees (for reviews, see Martin and Lindauer, 1977; Kirschvink and Gould, 1981; Frankel, 1984; Gould, 1984; Kirschvink, 1989). "  

"How does the magnetic field near the behavioral threshold ( ~ 1γ) compare with electromagnetic noise generated by nerves? After all, neurons in the brain are firing action potentials all the time, and these time-varying electric fields produce their own magnetic flux. As a starting point, we could use the law of Biot and Savart for the field produced at a radius, r, around a wire carrying current, I: B = mu₀  I/ 2 pi r, here mu₀ is the magnetic susceptibility of the vacuum (in S.I. units). Modeling a nerve fiber as a wire, we choose I = 10 nA and r = 10 pm. This gives B = 0.2 nT = 0.2 γ. This background field strength is not very much smaller than the behavioral thresholds. Empirically, magnetoencephalographs, which use SQUID magnetometers to measure external electromagnetic fields produced by nervous activity in the brain, register magnetic signals ranging from picotesla up to nanotesla."

"
Bialek, W. 1987.
Physical limits to sensation and perception. Annual Review of Biophysics and Biophysical Chemistry. 16:455-478.

Bullock, T. H., and W. Heiligenberg. 1986.
Electroreception. John Wiley & Sons, Inc., New York. 722 pp.

Heiligenberg, W. 1984.
The electric sense of weakly electric fish. Annual Review of Physiology. 46:561-583.

Kalmijn, A. D. 1982.
Electric and magnetic field detection in elasmobranch fishes. Science. 218:916-918."