Our neurobiological findings in monkeys have raised the possibility that, like occipitotemporal visual areas, superior temporal auditory areas send highly processed stimulus quality information to downstream targets via a multisynaptic corticocortical pathway. Previous evidence indicated that the auditory core (areas A1, R, and RT) on the supratemporal plane (STP) constitutes the first stage of cortical processing, with a serial progression from core outward, first to the belt and then to the parabelt. Our new data suggest that there is also a stepwise serial projection from A1 to R to RT and from there to the rostrotemporal polar field RTp. We have also investigated the subcortical connections of these auditory cortical areas and found that whereas A1 receives a preponderance of thalamic input from the ventral division of the medial geniculate (MGv), R receives a mix of inputs from both the ventral and dorsal subdivisions (MGv and MGd, respectively), and inputs to RT skew more heavily towards MGd. RT and RTp also project to the MGd and MGv. Injections beyond the core in RTp, RPB, and STGr revealed thalamic inputs from outside the medial geniculate body, including the suprageniculate nucleus and medial pulvinar. Additional projections were found to the striatum, with RT and RTp projecting to the tail of the caudate nucleus and the adjacent ventral putamen. RTp also projects to the ventromedial region of the head of the caudate nucleus and nucleus accumbens. Finally, RT and RTp are strongly interconnected with the amygdala. In summary, auditory fields rostral to the core receive little input from the auditory thalamus, suggesting that their physiological responses to sound are mediated by the previously described corticocortical pathways. The subcortical connectivity establishes circuits by which auditory processing may interact with systems involved in learning, attention, reward, and emotion. One of the fundamental features of sensory cortex is that it topologically maps the physical attributes of sensory stimuli. In the macaque auditory cortex, the attribute mapped is sound frequency, yielding tonotopic maps in the core regions on the STP. While neural responses in these areas have been studied in detail, the spatiotemporal activation of these maps by acoustic stimuli has not. We designed a micro-electrocorticographic (ECoG) array to record field potentials simultaneously from multiple areas of macaque auditory cortex. We chronically implanted four such arrays on the surface of the STP as well as the caudal superior temporal gyrus. The arrays allowed us to record auditory evoked potentials from a large expanse of auditory cortex simultaneously with high temporal resolution, while the monkeys passively listened to 180 different pure tone stimuli. First we examined the auditory frequency tuning of auditory evoked potentials at conventional field potential frequency ranges. This analysis revealed tonotopic maps that reversed frequency direction at putative areal boundaries. Although all field potential frequency bands produced similar areal boundaries, the smoothest tonotopic-reversal map was obtained from the high-gamma band of the evoked activity. Next, we estimated when each site showed significant discrimination among different stimulus frequencies by evaluating the high-gamma power in a 40 ms moving window. We found that the onset time of the discrimination increased along the caudal-to-rostral as well as the medial-to-lateral axes, consistent with the hypothesis that auditory information is serially processed in these two directions in parallel. As part of our investigation of auditory processing we trained monkeys on a task designed to assess auditory recognition memory. The task is comparable to those used regularly to test visual recognition. However, in contrast to the ease with which monkeys acquire the rule for delayed match-to-sample in vision, they strugggled for over a year to learn the same rule in audition. Further, once they learned the auditory task, their memory performance was (i) limited to short-term memory, and (ii) unaffected by lesions of the rhinal cortex;this is in sharp contrast to their memory performance in vision which (i) extends to long-term memory (ii) is severely disrupted by rhinal lesions. In a follow-up experiment to further determine the capacity, specificity, and duration of the maintained representation of an auditory stimulus, monkeys were trained on a serial delayed-match-to sample task analogous to one used previously to study these aspects of memory in vision. Performance was accurate in the absence of distracters, but degraded severely after the presentation of intervening nonmatch stimuli. Manipulation of the inter-stimulus interval confirmed that the performance degradation was attributable to the appearance of the intervening stimuli, not simply to the decay of memory over time. Analysis of performance by stimulus type showed a weak and counter-intuitive effect of sound category that favored simple stimuli (e.g., tones and noise) over species-specific vocalizations. These results suggest a strong effect of retroactive interference, such that the nonmatch stimulus disrupted the trace of the sample, effectively lowering the similarity criterion at which monkeys indicated a match. In short, we found that the auditory memory trace in nonhuman primates is limited to one or two items, is extremely fragile, and is highly vulnerable to overwriting by subsequent sounds. The impoverished auditory memory ability in monkeys contrasts not only with their excellent memory in vision but also with the human facility to encode auditory stimuli in LTM, thus raising the question of whether the human ability is supported in some way by speech and language. To test this possibility, we asked whether humans can store representations of speech sounds that can be neither repeated nor labeled (e.g., speech sounds played backwards). Our results indicate that the less that articulation and verbal labeling can be used to support storage of auditory information in LTM, the poorer the memory performance. This in turn has led us to propose that human speech and human auditory memory evolved together, possibly as a result of the evolution of the arcuate fasciculus from a primitive connection between the auditory and oromotor sytems present in nonhuman primates to the dense and complex linkage in humans, resulting in the elaboration of a Wernickes area as part of the auditory system connected to a Brocas area as part of the oromotor system. We have also begun assessing auditory memory in the three-generation KE family, half of whose the members have an inherited speech-language disorder, characterized as a verbal and orofacial dyspraxia. The core phenotype of the affected KE members (aKE) is a deficit in repeating words, especially non-words, and in moving the orofacial musculature. It is not clear whether the speech deficit results from combined working memory (WM) and motor output problems, or from motor output difficulties alone. We assessed WM using a test battery based on the Baddeley and Hitch WM model. The model posits that the central executive (CE), important for planning and manipulating information, works in conjunction with two modality-specific components: The phonological loop (PL), and the visuospatial sketchpad (VSSP). Our preliminary results indicate that the aKE, who have both structural and functional abnormalities in Broca's and Wernicke's areas, have a selective impairment in phonological WM, with relatively preserved CE and VSSP functions. Given that basic auditory sensory processes in the affected family members were found to be normal, we now suspect that auditory their WM may be impaired because of their structural and functional abnormalities at both ends of the arcuate fascicul
Showing the most recent 10 out of 20 publications
