Reductions in neural activity underlie behavioral components of repetition priming Gagan S Wig, Scott T Grafton, Kathryn E Demos & William M Kelley Repetition priming is a nonconscious form of memory that is accompanied by reductions in neural activity when an experience is repeated. To date, however, there is no direct evidence that these neural reductions underlie the behavioral advantage afforded to repeated material. Here we demonstrate a causal linkage between neural and behavioral priming in humans. fMRI (functional magnetic resonance imaging) was used in combination with transcranial magnetic stimulation (TMS) to target and disrupt activity in the left frontal cortex during repeated classification of objects. Left-frontal TMS disrupted both the neural and behavioral markers of priming. Neural priming in early sensory regions was unaffected by left-frontal TMS—a finding that provides evidence for separable conceptual and perceptual components of priming. Recent experience with an item leads to quicker recognition and classification of that item upon subsequent encounters. This implicit form of memory is commonly referred to as ‘‘behavioral priming’’ and occurs even in the absence of conscious remembering 1,2 . Neuroscien- tific investigations consistently reveal reductions in neural activity that accompany this repetition-based learning facilitation. These activity reductions are seen both at the level of single cells in nonhuman primates 3–5 and across a host of brain areas extending from posterior sensory to frontal cortices in humans 6–10 . Although this neural phe- nomenon generalizes across a range of behavioral priming situations, the loci of such ‘‘neural priming’’ effects vary and are restricted to a subset of the brain regions engaged during task performance with novel material. Repeated semantic classification of visually presented objects, for example, consistently yields reduced activity in extrastriate visual regions and in the left inferior frontal gyrus (LIFG) 11,12 . One speculation is that neural priming reflects fine-tuning of the neuronal response, or a suppression of neurons within the neuronal population, perhaps because those neurons that are no longer needed drop out of the responsive pool 3,13,14 . Such effects could occur either early in the processing stream at the level of object recognition in sensory cortices (perceptual priming) or at later stages during semantic classification in frontal and temporal cortices (conceptual priming) 15 . The precise neurophysiological mechanism that reduces neural activity is unclear, and the relationship between neural priming and behavioral priming remains indirect—evidenced only by the co-occurrence of these two phenomena. A fundamental question remains: does neural priming in a given brain region contribute to the behavioral facilitation afforded to repeated items, or is it epiphenomenal to behavior? One possibility is that neural priming in brain regions thought to be involved in conceptual priming (e.g., LIFG) is necessary for behavioral priming. Alternatively, neural priming in sensory cortices may subserve behavioral priming and the neural reductions observed in frontal regions may simply reflect a feed-forward propagation of the changes in neural activity arising earlier in perceptual regions. To test these possibilities, we combined fMRI and TMS to target and transiently disrupt left-frontal activity during an object classification task (Fig. 1). In an initial fMRI session, subjects performed a semantic classifica- tion task (living/nonliving) for a series of objects that were either repeated or novel. In a second session, subjects received TMS while performing the classification task on a new set of objects. The use of TMS allowed for noninvasive and transient disruption of cortical activity in a circumscribed region of cortex. We used activation maps from the initial fMRI session, which compared trials with novel objects to those with repeated objects, to identify subject-specific neural priming foci within the LIFG. These single-subject activation maps were then superimposed on the subject’s anatomical brain image and used to guide the positioning of the TMS coil on the subject’s head. This approach permitted real-time, continuous monitoring of the coil position with respect to the site of interest. For each presented object, TMS was delivered to either the LIFG region identified during fMRI scanning (left-frontal TMS) or the hand region of left motor cortex (control-site TMS). The motor region was included as a control site to ensure that TMS effects were specific to left-frontal stimulation and not a property of global cortical disruption. During the TMS session, each object was presented twice and was accompanied by a 10-Hz train of stimulation lasting 500 ms. Onset of the TMS was tailored to each subject’s individual response profile from the initial fMRI session (Fig. 2; see Methods). To assess both the behavioral and neurophysiological consequences of TMS, we performed a second fMRI scan on each subject immediately after the TMS session. Critically, this post-TMS scanning session allowed behavioral responses to be recorded in the absence of potentially Published online 31 July 2005; doi:10.1038/nn1515 Department of Psychological and Brain Sciences, Center for Cognitive Neuroscience, Dartmouth College, Hanover, New Hampshire 03755, USA. Correspondence should be addressed to G.S.W. (gagan.wig@dartmouth.edu). 1228 VOLUME 8 [ NUMBER 9 [ SEPTEMBER 2005 NATURE NEUROSCIENCE ARTICLES © 2005 Nature Publishing Group http://www.nature.com/natureneuroscience