134 Cathexis Revisited Corticolimbic Resonance and the Adaptive Control of Memory a DON M. TUCKER b,c AND PHAN LUU b b Psychology Department, University of Oregon, Eugene, Oregon 97403, USA c Electrical Geodesics, Inc., Riverfront Research Park, Eugene, Oregon 97403, USA S everal issues that Freud (1895) addressed in his Project, a “psychology for neurol- ogists,” remain fundamental problems for a neural model of mental processes. By theorizing how neurons could represent experience and organize behavior adaptively, Freud confronted the problem of establishing memories in the same neural mechanism that must be immutable in response to perceptual events. That is, according to Freud there is an inherent inconsistency between the changes that must occur in the functions of neurons in order for memory traces to be formed and the stability that must also be present in order for perceptual processing to be unprejudiced. By proposing a neural architecture for motivating cognition and behavior adaptively, he attempted to under- stand how needs and values come to shape the structure of memory and behavior. Per- haps most importantly, by framing his neurophysiological model to address broad psy- chological questions, Freud proposed a specific theoretical approach for bringing the evolutionary insights of nineteenth-century biology to bear on practical questions of human behavior. In this paper we revisit Freud’s neurophysiological speculations in light of current thinking in cognitive neuroscience. For important aspects of Freud’s approach, such as the emphasis on adaptive and homeostatic mechanisms in regulating cognition, mod- ern neuropsychological theory could still be informed by Freud’s work. With the con- cept of cathexis, for example, Freud integrated the mechanisms of memory with those of motivation in a way that is quite relevant to understanding the current evidence on corticolimbic interaction in memory and emotion. The idea that emotional excitement may be mediated directly by mechanisms of neuronal excitement remains a direct and compelling hypothesis in light of the modern evidence on emotion, kindling, and neural plasticity in corticolimbic networks. For other aspects of Freud’s theorizing, such as his separation of perceptual net- works from memory networks, we think the modern evidence shows that Freud’s pro- posals were wrong. Memory and perception are very difficult to separate, and in fact they appear to be achieved within common networks in the cortical pathways linking sensory areas with limbic structures. Even here, however, it was an important insight to recognize the inherent incompatibility between perceptual stability and mnemonic plasticity. This incompatibility—now known as “catastrophic interference” or “the stability/plasticity dilemma”—has proven to be a key problem in computational ap- proaches to neural models in the last half of the century. In this paper we entertain the hypothesis that the stability–plasticity dilemma may not be restricted to neural network simulations. It may reflect a fundamental fact of c This work was supported by NIMH Research Grants MH42128 and MH42669 to the Uni- versity of Oregon and Small Business Innovation Research Grants MH50409 and MH51069 to Electrical Geodesics, Inc.