412 A. L. TOWE 



1958). Such a phenomenon can be called an inhibitory interaction in the 

 general sense of the definition used in this symposium. 



This inhibitory interaction could result from either of two mechanisms. 

 The conditioning input might discharge some peripheral neurons also 

 discharged from the test field, leaving them refractory or less excitable to the 

 subsequent testing input. This occurrence would reduce the size of the 

 excitatory input aroused by the testing stimulus; the cortical unit might then 

 fail to discharge at all. On the other hand, the conditioning input might 

 directly inhibit some neurons that are ordinarily excited by the testing input, 

 reducing the size of tiie subsequent test excitatory volley. In order to discover 

 which mechanism prevails, it is necessary to observe more than the response 

 probability changes. Figure 2 shows what happens to the initial discharge 

 latency during the interaction. Two distinct changes occur. When the condi- 

 tioning stimulus is applied well inside the unit's excitatory receptive field, 

 the latency of the test response clearly and systematically shortens. When the 

 conditioning input is appUed on the edge of, or just outside, the unit excitatory 

 receptive field, the latency of the test response lengthens. These two response 

 patterns cannot be dismissed as "random", for they are statistically highly 

 reliable. 



We are already famihar with unit response variations of this nature (Ken- 

 nedy and Towe, 1958). The initial spike latency of a postsynaptic unit varies 



A ADJACENT 

 1 



(msec) . 

 12- 



X" 



B NON- ADJACENT 



Fig. 2. Two patterns of interaction from two different cortical units, a. Unit 

 from 644 /< down in p.c.g. Conditioning digit V fired unit with probability (/?) 

 of 08. Testing digit IV fired unit with p = 0-62 after 15-2 msec mean latency. 

 In first 15 msec of interaction the test response latency (/) progressively shortened 

 to more than eight standard deviations (s) below the unconditioned testing Z; 

 L was also shorter during recovery, b. Unit from 1829 /< down in p.c.g. Condi- 

 tioning digit IV never fired unit. Testing digit II fired unit with /? = 082 after 

 14- msec latency. Throughout interaction, L of the testing response was greater 

 than 14-0 msec, increasing by more than three standard deviations (s) above the 

 unconditioned testing L. 



