Among the many contributions of the Montreal Neurological Institute (MNI) to the surgical treatment of refractory epilepsy, the recognition by Wilder Penfield that awake neurosurgery for epilepsy provided a unique opportunity to investigate the cortical organization of human cognition has been arguably the most significant, (Penfield & Jasper, 1954; Penfield & Roberts, 1959; Penfield & Perot, 1963). Among Penfield’s residents in the early 1940s was Arthur Ward, Jr., who brought to the MNI a background in physiology acquired in another classic venue, the Yale primate laboratories of the late 1930s. Dr. Ward was the author’s mentor, and from him came both traditions: epilepsy surgery as an opportunity to study human cognition with the stimulation localization technique of Penfield and Jasper, and the single neuronal recording techniques of the primate neurophysiologists. This article reviews some of the author’s investigations of the neurophysiologic correlates of language, recent verbal memory and verbal learning in this setting, and consenting patients undergoing temporal lobe resections with an awake surgical technique (Ojemann, 1995). Those investigations have utilized several intraoperative techniques, including electrical stimulation mapping, optical imaging of the “intrinsic signal” (Haglund et al., 1992), and extracellular recording of single neuronal activity, each providing different perspectives on these cognitive processes. Stimulation mapping links a brain region to a cognitive process by interfering with that process. For language and memory, there is evidence that this effect predicts the effects of resection of that tissue (Ojemann, 1983; Ojemann & Dodrill, 1987; Haglund et al., 1994a), presumably then the activity of the tissue where interference is evoked is crucial for the behavior, at least at that point in time. By contrast, optical imaging and recording of single neuron activity are physiologic changes correlated with the behavior, indicating where the physiologic activity is occurring, but not necessarily in tissue that is crucial for it. These investigations represent collaboration between the author and many associates from other disciplines, particularly neurophysiology and neuropsychology. Classical lesion effects have established that the temporal lobe of the dominant hemisphere contains structures essential for language, verbal memory, and verbal learning. Within dominant temporal lobe, language functions are usually related to lateral neocortex, particularly in posterior–superior temporal lobe, whereas recent verbal memory and learning effects are usually related to medial temporal structures, particularly hippocampus. Electrical stimulation mapping of lateral temporal neocortex of the dominant hemisphere also commonly produces interference in language measures. However, the sites of interference are often focal areas of 1 cm2 or so, considerably smaller than the classic posterior temporal language area. There is considerable individual variation in the location of these focal areas (Ojemann et al., 1989). This includes extension into area thought to be uninvolved in language on anatomic criteria, such as anterior portions of superior and middle temporal gyri in some patients, whereas in others, areas usually considered important for language such as posterior superior temporal gyrus (“Wernicke’s area”) are spared. Some of this individual variation differed related to patients’ preoperative verbal abilities as measured by verbal IQ, and to gender. Language localization also differed between children and adults (Ojemann et al., 2003). Temporal cortical stimulation frequently interferes with different language measures at different sites, including separate sites for naming in two languages (Ojemann & Whitaker, 1978; Lucas et al., 2004) and naming compared to reading (Ojemann, 1989). Somewhat surprisingly, stimulation of lateral temporal neocortex has also interfered with performance on a recent verbal memory measure, particularly when the current is applied during encoding or storage of the verbal item. These sites have often been separate from those for which stimulation interferes with naming of the same items, so that there appears to be separation of the crucial temporal cortical sites for recent episodic verbal memory from those for language, even though the recent memory measure involved overt production of the name of the item to be encoded in memory (Ojemann, 1978; Ojemann & Dodrill, 1985). Sites with interference in the memory measure were particularly likely in anterior temporal lateral neocortex. Resection of the sites with stimulation interference during the memory measure was associated with an increased postoperative verbal memory deficit (Ojemann & Creutzfeldt, 1987; Ojemann & Dodrill, 1987). In contrast to these lateral temporal cortical effects on recent verbal memory and to the classical effects of medial temporal lesions on memory, it has been difficult to show medial temporal or hippocampal stimulation interference on recent memory independent of evoking seizures (Ojemann & Creutzfeldt, 1987). Stimulation effects have also been compared between novel and overlearned items (Ojemann et al., 2002b). Interference was evoked from a wider area for novel items than for overlearned items, suggesting that the regions of brain crucial for learning are more extensive than those crucial to execution of a learned task. Optical imaging during language measures has shown changes in dominant temporal neocortex, but with a distribution wider than the sites of stimulation interference on the same language measure in the same subject (Haglund et al., 1992). In an unpublished case study, optical imaging changes in dominant temporal neocortex were also evident during recent memory retrieval of encoded names, but in a pattern different from that for naming in the absence of the instruction to remember the name. As with the stimulation mapping findings, optical imaging changes during memory retrieval spared sites where stimulation had interfered with naming, but showed more extensive involvement of anterior temporal neocortex (M. Haglund, D. Hochman, and G. Ojemann, unpublished data). Extracellular recording of changes in single neuronal activity in lateral temporal cortex that correlate with language measures has shown substantial differences from the findings with stimulation mapping, likely reflecting the difference between a technique that shows where neurons are active and participating in a behavior from those regions that are essential for that behavior. Changes in single neuronal activity, compared to activity during control behaviors, occurred in equal proportions of neurons from dominant or nondominant lateral temporal cortex for auditory word perception and repetition, visual object naming, and word reading (Creutzfeldt et al., 1989a,b; Schwartz et al., 1996; Ojemann & Schoenfield-McNeill, 1999). Because dominance for language had been established in all subjects of these investigations, based on intracarotid amobarbital perfusion assessment (Wada test), changes in single neuron activity during the language tasks that lateralized to the dominant hemisphere were sought (Schwartz et al., 1996). Two changes were identified. Changes in neuron activity can be relative increases (excitation) or relative decreases (inhibition) compared to the control measures. Relative inhibition during naming was one feature significantly lateralized to dominant hemisphere recordings. This relative inhibition may represent an inhibitory surround of a focal area of excitation elsewhere in temporal cortex, or perhaps the relation between temporal association cortex activity and subcortical activity is similar to that between motor cortex activity and spinal motor neuron activity, where the cortex modulates the greater subcortical excitation by changing the degree of inhibition. In a study of lateralized differences in neuron activity during a visuospatial task, inhibition was observed but lateralized to the nondominant hemisphere (Lucas et al., 2003), suggesting that the relative inhibition is a feature of dominance rather than specific to language tasks. The other language change lateralized to the dominant hemisphere was earlier activity. By contrast, nondominant activity changes with language tended to be excitation late, at the time of the speech output that was part of all the language tasks. When several different language measures were administered during recording from the same neuron, the most common pattern was changes with only one task. This included changes during object naming compared to word reading (Schwartz et al., 1996) and changes with naming in only one of two languages (Ojemann, 1990). Most of the changes in activity during language measures have been in the frequency of activity. However, recordings from a few neurons have had patterns of activity that appeared to be specific to specific words or to prosody (Ojemann et al., 1988; Creutzfeldt et al., 1989b). The same recent verbal memory paradigm was utilized for stimulation and single neuron studies. Neuronal activity during the encoding stage was compared to that during identification of similar items, but without the instruction to remember them. Therefore, the two tasks differ only in that instruction, to remember the item or not. This instruction changed the frequency of activity in a large proportion of temporal neocortical neurons. In recordings from 239 neurons in 86 subjects, activity was significantly altered in 135 (57%), (Ojemann et al., 1988; Haglund et al., 1994b; Weber & Ojemann, 1995; Ojemann & Schoenfield-McNeill, 1998, 1999; Ojemann et al., 2002a, 2009). As with language measures, the proportion of neurons changing activity with recent memory encoding was similar in dominant and nondominant temporal lobes. Changes in activity during encoding were present throughout lateral temporal neocortex. When divided into that occurring early, during perception and processing of the items to be encoded, late, related to the overt speech output associated with encoding, or sustained throughout encoding (Ojemann et al., 2009), early perceptual changes occurred in superior temporal gyrus and later processing and output changes in middle gyrus. These changes are specific to recent memory, and not to identification without the memory instruction. Activity sustained throughout encoding was widely recorded from lateral cortex. However, that recorded from the superior and middle thirds of middle temporal gyrus was significantly more likely to represent relative inhibition (compared to identification without the memory instruction) than that recorded from surrounding cortex. Relative inhibition during speech output was present in recordings from a portion of this same middle temporal gyrus but in different neurons. About 10% of the 98 neurons included in this study of timing of activity during encoding had a pattern suggesting a simultaneous convergence of sustained tonic activity, perhaps representing an attentional effect specific to the task but not the item, and phasic activity during early perception and processing, more likely item specific. Simultaneous convergence of inputs to a neuron has been proposed as a mechanism for memory with potentiation of synaptic inputs to a neuron (Hebb, 1949). Learning was assessed with a word pair association paradigm, where the same 20 words (concrete nouns) were used under three conditions—identification of the words, as the items to be encoded in recent memory, and as 10 unrelated word pairs. The word pairs were presented in a series of trials, each trial including a presentation of each pair, which the subject read aloud, and a test of learning of each pair, with the first word presented, which the subject read and then gave the second word, if learned. These measures have been used in two studies (Weber & Ojemann, 1995; Ojemann & Schoenfield-McNeill, 1998). In one study, activity was significantly changed by word reading in 36% of neurons, by memory encoding in 77%, and by the learning task in 100% of the recorded neurons. Changes in activity that discriminated learned from unlearned pairs on the first presentation trial were identified. These changes occurred in neurons that were inhibited by the word reading task but excited by memory encoding. The activity change that discriminated learned from unlearned pairs in these neurons was significantly increased activity sustained during and after the word pair was correctly read on that first presentation trial (Ojemann & Schoenfield-McNeill, 1998). Whether this sustained activity represents continued rehearsal of the pairs that are learned, or reflects tonic activity associated with attentional mechanism is unknown. Once the association was learned, the level of activity began to decrease, beginning with the second trial after the pair was learned (Weber & Ojemann, 1995). Subjects who learned the pairs readily had significantly greater activity in neurons also related to overt word reading than did subjects who learned words poorly. Conversely the early learners had less activity in neurons unrelated to identification or memory than the poor learners. Therefore, these neuronal events seem to be essential for learning. Based on these studies the neural events in temporal cortex during verbal associative learning have been modeled as a transient sustained increase in activity during encoding of the association, activity sustained after identification of the pair, but rapidly declining within a few trials after initial learning, so that later retrieval of the association requires activity in many fewer neurons. Indeed there is the suggestion that for overlearned items, some of those neurons may be actively inhibited. Note that these findings with single neuron recording of less activity for overlearned items are similar to the findings with stimulation mapping indicated earlier, of smaller crucial areas for overlearned items. The author has no conflicts of interest to declare.