Cortical Mechanisms Of Auditory Perception And Memory

National Institute Of Mental Health

Investigators

Linked publications, trials & patents

Abstract

(1)The cortical auditory system is far more extensive than expected, as it includes not only the entire superior temporal gyrus but also large portions of the parietal, prefrontal, and limbic lobes. Several of these areas overlap with previously identified visual areas, suggesting that the auditory system, like the visual, contains separate pathways for processing stimulus quality and stimulus location. (2) This division between stimulus quality and stimulus location processing was demonstrated for auditory working memory. Human subjects were asked to remember the identity of voices or of their location in an fMRI study. (3) Monkeys, like humans, appear to use the auditory system of the left hemisphere preferentially to process vocalizations. This functional asymmetry was demonstrated in monkeys by comparing metabolic rates, using FDG and PET scanning while the animals listened to conspecific calls compared with numerous other sound classes. (4) New experiments using high-field fMRI with a new method that allows us to combine and analyze fMRI data from multiple sessions we have showed not only tonotopic maps and the reversal in frequency representation at the junction of A1 and R but also demonstrated a correlation in the increase BOLD response to three different sound levels (70, 80, and 90 dBSPL) in both high and low frequency tones. (5) Monkeys trained on a task designed to assess auditory recognition memory, were impaired after removal of either the rostral superior temporal gyrus (rSTG) or the medial temporal lobe (MTL) but were unaffected by lesions of the rhinal cortex (Rh). These results compared with those obtained in other sensory modalities, in which Rh lesions produce severe impairment, lead to the tentative conclusion that the monkeys in this auditory study were unimpaired after Rh lesions because they had performed the task utilizing working memory. A normal monkey performing our version of auditory DMS resembles a monkey with Rh cortical lesions performing visual DMS; neither seems able to store long-term stimulus representations. This inability suggests that the auditory stimuli were processed without the participation of the Rh cortices and consequently ablating the Rh cortices had no deleterious effect. The correlate of this proposal is that, like the residual mnemonic ability in vision after Rh lesions, the mnemonic ability in audition observed here in normal monkeys rests entirely on mechanisms mediating short-term or working memory. This proposal that performance on our auditory DMS task was based on working memory implies it is the form of memory that impaired after both the rSTG and MT lesions. This implication after rSTG lesions is consistent with the anatomical position occupied by rSTG in the cortical auditory system. A series of anatomical studies suggests that rSTG may be a late station in the ventral processing stream for audition much like area TE is for vision, and so it is highly likely that rSTG serves an important role in short-term auditory memory just as area TE does in short-term visual memory. The impairment produced by the MT ablation may have been a result from collateral damage and not hippocampal damage, such as severing the projections of rSTG to downstream areas in the prefrontal cortex, medial thalamus, or both. (6) An important aspect of auditory processing is the spectral and temporal integration of acoustic information. We recorded neuronal responses to (a) pure tones and bandpassed noise in order to obtain frequency tuning curves, (b) ripple stimuli in order to generate spectrotemporal receptive fields, and (c) linear frequency modulated stimuli, which are natural components of monkey calls. Cortical neurons showed a variety of temporally structured types of response, including tonic, pauser, and offset responses, which appear to have different distributions in the two primary auditory areas, A1 and R. Although most of the neurons respond preferentially to a single best frequency, some neurons show multiple tuning peaks, suggesting that these neurons integrate spectral information over a wide range of frequencies, which may be important in mediating auditory pattern recognition. Also, 75% of A1 and R neurons were selective for FM direction and/or rate, with response distributions suggesting that rate and directionality are independently represented in monkey primary auditory cortex. (7) Neurons in the rostromedial portion of the supratemporal plane (rSTP, extending 3-11 mm caudal to the temporal pole) respond to a variety of sounds, including monkey-vocalizations. We have now extended the recording caudally to primary auditory cortex (area A1) to permit direct comparison of neuronal properties in rostral and caudal stations of the putative stimulus-quality processing stream in response to the identical set of sounds (20 monkey vocalizations and 20 other auditory stimuli). The cells in the two areas showed many similarities. For example, the proportion of cells responding to sounds in a given category was comparable in the two areas (A1 vs. rSTP: 10% vs. 15%, monkey vocalizations only; 19% vs. 20%, other auditory stimuli only; 22% vs. 13%, both; 19% vs. 8%, white noise only; 38% vs. 44%, non-responsive). Also, cells in both areas showed suppression as well as excitation, with some cells in each area exhibiting both of these response patterns to different sounds, with excitatory responses in the two areas often showing both an early and a later peak. In addition, in the case of excitation, the onset latencies in both areas were correlated with the early peak magnitudes (i.e. the shorter the latency, the stronger the response). However, the number of sounds that evoked a significant auditory response in individual cells was significantly smaller in rSTP than in A1, the mean latency of the response in rSTP was delayed by 50 ms compared to that in A1, and, in the case of excitation, the percent increase in average spike rate of the early peak relative to the spontaneous firing rate (baseline) was only 85% in rSTP compared to 320% in A1. Also, the variance in both latency and response magnitude between early and late peaks was greater in rSTP than in A1. A sliding-window analysis revealed that in both areas the late peak responded to fewer sounds than the early peak, implying a sharpening of stimulus selectivity in the later phase of the response in individual cells. Together the results suggest that although both A1 and rSTP neurons have many properties in common, supporting the proposal that they both form part of a stimulus-quality processing stream, rSTP is a late component of the stream and has a greater degree of stimulus selectivity than A1. (8) At the level of the A1 and R, the firing rate of a neuron may track periodic features (e.g. frequency or amplitude modulation), but such tracking is not apparent in neurons rostral to the core auditory fields. In a separate experiment we recorded spike trains of single neurons in rSTG to determine if discharge patterns carry stimulus specific information. Monkeys performing a short-term memory task were presented with a stimulus set of 21 sounds. For each of the 21 stimuli, response coefficients were averaged across repeated presentations to determine if temporal characteristics of the response were consistent within stimuli. Compare to randomly drawn bootstrap samples the first principle component (PC) of nearly all tested units showed a significant effect of stimulus identity. A simple classifier based on waveform shapes suggests that mutual information between response waveform and stimulus identity increases as up to 20 PCs are used. These results suggest that in higher-order sensory cortices, temporal aspects of spike trains may be more important for representing dynamic auditory stimuli than they are for representing static visual images.