R. W. GERARD 27 



localization is, of course, present, as shown by the aphasic defects with 

 regional lesions and by particularized recollections inducec^ by local 

 temporal stimulations. But even these are hardly cell by cell; and it is far 

 more probable that large numbers of neurones, in assemblies or masses, in 

 different patterns and other orderings, are involved in each memory. 

 Partial engrams — of percepts, images and acts — are built into larger ones 

 — concepts, imaginings and skills — much as a small assortment of amino- 

 acids is used to build a limitless variety of proteins. In the same way, learn- 

 ing goes from letters to words to sentences, with plateaux of achievement 

 at each larger unit; and bits of information become aggregated into larger 

 'chunks' so that a greater quantity can be handled in a given time (Miller, 

 1956). Not only spacial relations, but also temporal ones must be proper, 

 witness the great disturbances to thought and speech produced by delayed 

 auditory feed-back. 



Whatever the micro-locus of the memory trace, most learning involves 

 the cortex. Besides the evidence of cortical localization by stimulation and 

 lesion, there is the further evidence of a general parallel between learning 

 and memory capacity on the one hand and general cortical size on the 

 other, and also there are the recent psycho-physiological experiments 

 initiated by Sperry (1959). With the optic chiasm cut, so that incoming 

 messages from each eye reach only the ipsilatcral hemisphere, the two 

 hemispheres remain connected primarily through the corpus callosum. 

 If conditioning is carried on with cMie eye, a correct response can be elicited 

 through either eye so long as the callosum is intact; but after this is also 

 cut, only the eye used in training can elicit the learned response. The 

 engram, while available to both hemispheres when these remain anatomi- 

 cally connected, is as clearly localized in only one. Comparable tmdings 

 have been made on learning set or learning to learn. Other evidence for a 

 cortical engram, and one which acts as a template for further engram 

 development, comes from the work of Meyer (1958). Removal of both 

 occipital lobes, but with a two-week interval between ablations, leaves a 

 rat with pattern vision essentially intact if ordinary visual experience is 

 possible during the interval; but if the animal experiences no pattern 

 vision between the removal of the hrst lobe and that of the second, a 

 pernianent loss of pattern vision results. The endurance for months of 

 figural after-effects in the Kohler's satiation-illusion (Wertheimer and 

 Leventhal, 1958) further indicates a cortical locus of the engram. 



Not all fixation, however, is in the cortex. Although ordinary condi- 

 tioning of the spinal animal remains highly dubious, there is solid evidence 

 of card changes produced in the intact animal. If one cerebellar hemisphere 



