2 INVERTEBRATE PHYSIOLOGY 



Now, what is the whole which is referred to? At the levels below be- 

 havior, neurophysiologists today regard it as a pattern in time and space of 

 quantal events, each event brief compared to the events of behavior ; these 

 are the impulses in the efferent nerve fibers. At the level of the single 

 neuron we may perhaps best express the whole as the probability of firing 

 within the next given interval of time, or we may revise "firing" to read 

 "change of state influencing another neuron," since it seems to me im- 

 portant to recognize the possibility of subthreshold events as adequate 

 stimuli even though clear cases are not yet known. We can formulate our 

 definition in simple terms like these if we but recognize, and then put aside 

 for the present, those variables which the neuron integrates into its proba- 

 bility of firing which are not immediately determined by other neurons, e.g., 

 general chemical milieu, physical deformation, and temperature. Expressed 

 in terms of impulses or changes of state effective upon other neurons, in- 

 tegration at the unit level then becomes in the general case a relation be- 

 tween input and output which is either more or less than one. Usually this 

 means the algebraic summing of separate neuronal channels one or more of 

 which produces more or less than one output pulse for each input pulse ; so 

 long as this is true, the channels may have equal or different weights in 

 their effect upon output and the same or opposite sign, i.e., excitation or 

 inhibition. But we have to admit also the case where only one input channel 

 reaches the neuron under consideration, for in our type of system a quali- 

 tative difference between inputs is not an essential condition ; the inte- 

 grating cell does not know whether the signals come in the same or differ- 

 ent channels. The essential feature is that the neuron place some value, 

 other than one, upon at least some of its incoming signals, according to 

 their intensity, time course, time of arrival, and the locus upon the neuron 

 where they impinge. This definition of integration at the neuron level will 

 then include all junctions except those that are purely 1 :1 relays. It will 

 certainly include many neuroeffector junctions in which therefore the 

 nonnervous cell is the integrating cell. Sensory neurons certainly integrate 

 in the broad meaning given first, putting together different quantities in 

 the milieu into a probability of firing. And they may do this in part by means 

 or with properties which will help us to illuminate junctions. But if any 

 should object to the notion that receptors already integrate, they may wish 

 to exclude receptors on the ground that there is no input from other neurons 

 — it is not nervous integration. But sometimes there is ! The same cell, the 

 same terminal ramifications may be transducers of mechanical simuli and 

 postsynaptic elements under nervous control (e.g., Kuffler and Eyza- 

 guirre, 1955; Lowenstein, 1956). 



You have patiently listened to my definition of integration. We are 

 supposed to talk about recent advances in invertebrates, and I must accord- 



