Principles of Stimulus Coding 11 



large changes in stimulus energy would, in a linear transforming 

 system, be signalled only by very small changes in frequency 

 of impulses. A hypothetical photosensory neuron which was 

 sensitive to light energies over the entire available range (lo^®) 

 would, in response to an increase of light intensity by a factor of 

 lo', increase its firing frequency by only one impulse per second. 

 Such small frequency increments would possibly augment, in a 

 statistical manner, the detectable input from large numbers of 

 individual sensory cells; however, it is extremely unlikely that 

 changes of this magnitude could be correctly interpreted by 

 the higher nervous centers on a single neuronal basis; for 

 random fluctuations in events at central synapses probably 

 exceed in amplitude any individual response increment of one 

 part in lo'. 



One obvious way to circumvent the apparent incompatibility 

 of high sensitivity to stimulus energies and the broad dynamic 

 range of the nerve cells involved would be the provision of several 

 different populations of primary neurons or sensory cells. In 

 such an arrangement each population would respond to the same 

 stimulus modality or quality, but collectively they would have 

 dynamic ranges which would overlap and would vary in sensi- 

 tivity over several orders of magnitude. In fact, there is adequate 

 evidence that such a principle of functional division among sensory 

 cells does operate in many animals. The acoustic organs of 

 Noctuid moths provide an exceptionally striking example. 

 Sensory neurons in these organs are capable of responding to 

 sounds over a broad frequency spectrum in the ultrasonic range. 

 These cells are, in fact, able to detect the sounds emitted by 

 predacious bats, and their signals influence the behavior of the 

 moth when stimuli of this nature are encountered. It is, perhaps, 

 surprising that the detection of such compressional energy 

 appears to be accomplished by only two pairs of primary sensory 

 neurons — one pair innervating the acoustic organ on either side 

 of the animal. Although both auditory cells of a pair respond to 

 the same broad frequency range of sound, one cell on each side 

 is more sensitive (by about 20 decibels) than its partner.^® Thus, 

 as the intensity of an artificially generated ultrasonic pulse is 

 increased — or as a hunting bat approaches the moth — the fre- 

 quency and number of impulses in the more sensitive neurons 



