582 BRAIN MECHANISMS AND LEARNING 



entorhinal activity, and during the actual approach typical bursts of 6 

 cyclcs/scc. activity appeared. On extinguishing the response, by removal of 

 the reinforcing food reward, all aspects of 6 cycles/sec. activity disappeared 

 during the period when approach was permitted. Some 4 cycles/sec. 

 activity occurred during the delay period. On retraining, the rapid return 

 of the extinguished response was accompanied by the reappearance of 

 typical 6 cycles/sec. activity in the entorhinal area. 



Now it is obvious that the subjective and essentially empirical inter- 

 pretation of EEG records is open to criticism, and is often quite uncon- 

 vincing. We have, therefore, endeavoured to overcome some of these 

 subjective aspects of our interpretation by transferring these training 

 records to IBM pimchcards. Although a tedious procedure, this per- 

 mitted the computing of various aspects of the 4 and 6 cycles/sec. wave 

 trains. We have been fortunate in enjoying the use of the 1BM-709 

 electronic computer, installed at the Western Data Processing Center, at 

 the University of California at Los Angeles. The records have been 

 subjected to both auto- and cross-correlation studies. Each hippocampal 

 lead can be auto-correlated; that is, compared with itself, in order to 

 determine aspects of inherent rhythmicity in the slow-wave bursts, and to 

 detect hidden rhythms. In cross-correlation, or comparison of one lead 

 with another, it is possible to determine the phase relations between the 

 slow waves of one hippocampal lead and another. 



It will be seen from Fig. 4 that the auto-correlation reveals a high degree 

 of rhythmicity, essentially sinusoidal, in the records from both dorsal 

 hippocampus and the entorhinal area, at approximately 6 cycles/sec. 



Cross-correlations of these records have revealed striking and consistent 

 differences in the phase relationships noted in early training when com- 

 pared with those from the same animal in late training. In early training, 

 the entorhinal lead lags behind, by some 20-30 msec, the 6 cycles/sec. 

 activity in both CA., and CA4 zones of the dorsal hippocampus. This is 

 precisely what would be predicted from acute experiments involving 

 direct septal stimulation. By contrast, cross-correlations in late training 

 show clearly that the initial phase arrangements arc now reversed. Now, 

 the entorhinal area consistently leads the discharge in the dorsal hippo- 

 campus by as much as 65 msec. 



We may, perhaps without undue licence, extrapolate these findings to 

 the morphological systems discussed in Fig. i. In the untrained animal, the 

 phase distribution in the 6 cycles/sec. bursts, associated with the approach 

 performance, is consistent with the system of pathways proposed by 

 Elliot Smith (19T0) and Herrick (1933), and as confirmed in our acute 



