Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
, 114 (2), 1158-71

Responses of Primate Frontal Cortex Neurons During Natural Vocal Communication

Affiliations

Responses of Primate Frontal Cortex Neurons During Natural Vocal Communication

Cory T Miller et al. J Neurophysiol.

Abstract

The role of primate frontal cortex in vocal communication and its significance in language evolution have a controversial history. While evidence indicates that vocalization processing occurs in ventrolateral prefrontal cortex neurons, vocal-motor activity has been conjectured to be primarily subcortical and suggestive of a distinctly different neural architecture from humans. Direct evidence of neural activity during natural vocal communication is limited, as previous studies were performed in chair-restrained animals. Here we recorded the activity of single neurons across multiple regions of prefrontal and premotor cortex while freely moving marmosets engaged in a natural vocal behavior known as antiphonal calling. Our aim was to test whether neurons in marmoset frontal cortex exhibited responses during vocal-signal processing and/or vocal-motor production in the context of active, natural communication. We observed motor-related changes in single neuron activity during vocal production, but relatively weak sensory responses for vocalization processing during this natural behavior. Vocal-motor responses occurred both prior to and during call production and were typically coupled to the timing of each vocalization pulse. Despite the relatively weak sensory responses a population classifier was able to distinguish between neural activity that occurred during presentations of vocalization stimuli that elicited an antiphonal response and those that did not. These findings are suggestive of the role that nonhuman primate frontal cortex neurons play in natural communication and provide an important foundation for more explicit tests of the functional contributions of these neocortical areas during vocal behaviors.

Keywords: antiphonal calling; marmosets; natural behavior; primate frontal cortex; vocal communication.

Figures

Fig. 1.
Fig. 1.
A: exemplar recording of an antiphonal calling exchange between two marmosets. The phee call shown on channel 1 (top) is the stimulus broadcast by the interactive playback software employed in this study. The phee call on channel 2 is the antiphonal call response produced by the subject. The schematic of the experimental setup is shown below. Subjects (red circle) were positioned on the opposite side of a test chamber from a speaker (black square). A cloth occluder was positioned in the middle of the room equidistant between the subject and speaker. A microphone (mic) was positioned in front of the subject. B: spectrogram (top) and amplitude waveform (bottom) of a representative 2-pulse phee call. C: schematic drawing of marmoset cortex and the position of the 4 microelectrode recording arrays in frontal cortex. The blue square indicates the location of the sections plotted in D. Cytoarchitectural boundaries in marmoset frontal cortex are shown (Paxinos et al. 2012). LS, lateral sulcus; STS, superior temporal sulcus. D: serial frontal cortex sections for monkey R01. A dashed line shows the location of the electrode array from rostral-caudal. A vertical line provides an orientation for dorsal (D) and ventral (V) orientations, while the rostral (R)-caudal (C) axis is shown along the edge of the stack of individual sections.
Fig. 2.
Fig. 2.
Representative tissue for 3 stains (CO, Nissl, vGluT2) used in histological reconstruction of recording sites. The CO section shows the different areas of prefrontal cortex on the lateral surface. A dashed line and/or red star indicates an electrode tract. Boundaries of prefrontal cortex areas are shown: 8aV, 8aD, 45, 47, ProM.
Fig. 3.
Fig. 3.
Representative exemplar neurons recorded in marmoset frontal cortex during vocal production. Four individual neurons are shown (A–D). A raster (top) and mean spike rate (bottom) is shown for each neuron. A schematic of a phee call is shown in the middle in gray. Spike timing is aligned with the onset of the 1st pulse of 2-pulse phee calls (0 s). The gray bars in the raster indicate the timing of each pulse of the phee call for each trial, while vertical black lines on each horizontal trial represent an action potential. The vertical red lines in the peristimulus time histogram (PSTH) plot indicate the mean timing of each pulse in the phee calls produced during the recording session. The mean spike rate is plotted in a solid black line, while ± 2 SE is plotted in the dashed black lines. 2 SE bounds were generated using a Jacknife procedure that takes different parts of the data, 90% in each draw with a different 10th of trials left out each time, and computes the mean and standard deviation (STD) over those different parts. It then estimates the mean as the mean of those means, and the standard error of the mean as the STD of the means times square root(N) where N is 10 in this case. An insert of the same unit is shown to the right, highlighting neural responses during the 2nd pulse of each phee call. Spike times are aligned with the 2nd phee call pulse (0 s). Gray bars indicate the timing of the 2nd pulse only; otherwise data are plotted identically to the raster/PSTH for the entire 2-pulse phee call. Above this insert is a schematic drawing marmoset frontal cortex with the location of the individual neuron marked with a red dot.
Fig. 4.
Fig. 4.
A: a schematic drawing of marmoset cortex with the characteristic responses at each electrode location in frontal cortex. Red circle, vocal signal-responsive neurons only; black circle, vocal motor-responsive neurons only; red circle with black outline, vocal signal- and vocal motor-responsive neurons at this location; white circle, single units recorded but no vocal signal- or vocal motor-responsive neurons; gray circle, no units above neural and behavioral thresholds. The boundaries of the different frontal cortex regions are shown with dashed lines. The outline of each electrode array is shown in blue. B: bar graph plots the % of neurons exhibiting significant changes in neural activity to vocal-signal stimuli (vSig, black bar), during vocal-motor production (vMtr, red bar) or to both vocal-signals and vocal-motor (black bar with red outline). C: coefficient of variation (CV) of spike rate for each vocal-signal (vSig)- and vocal-motor (vMTR)-responsive neuron is shown during both the respective “baseline” and “behavior” periods. Blue lines indicate the mean CV. Statistically significant differences are indicated; n.s., not statistically significant. D: response-index comparing spike rate between the antiphonal and independent contexts. Red bars plot responses for vocal signal-responsive neurons, while black bars plot responses for vocal motor-responsive neurons. A positive (+) index indicates a stronger response when the behavior component occurred during antiphonal calling (ant), while a negative (−) index indicates a stronger response when the behavior component occurred independently (ind).
Fig. 5.
Fig. 5.
Neural classification of phee calls. On the left side, median accuracy of logistic regression for single units to classify between vocal-signals and vocal-motor activity (A), vocal-signal processing between antiphonal and independent contexts (B), and vocal-motor production between antiphonal and independent contexts (C) using firing rates from 0 to 4 s from phee call onset for each category. Units are ordered from worst median performance to best. Error bars are 95% confidence intervals. The 10 single units weighted most by the median population classifier are highlighted in blue. On the right side, the distribution of median classification accuracy for the single units (gray) is compared with the distribution of population classifier accuracies (magenta) derived by a bootstrap procedure.
Fig. 6.
Fig. 6.
Summary of results for vocal motor-responsive neurons. A: plots the mean (±2 SE) for the population of neurons exhibiting excitation (top, blue) or suppression (bottom, green) during vocal-motor production. B: a schematic drawing of marmoset frontal cortex with the locations of vocal motor-responsive neurons. Each recording site at which these neurons were found are shown and marked by the type of responses characterizing that location. Green, suppressed; blue, excited; green circle/blue border, neurons exhibiting either suppression or excitation were found. C: latency to peak response for both suppressed (green) and excited (blue) vocal-motor neurons is shown in a scatter plot in the middle. Spikes were grouped into 100 ms bins. Histograms plot the # of neurons at each latency for suppressed (bottom) and excited (top) vocal-motor neurons. D: plots the CV for suppressed (green) and excited (blue) vocal-motor neurons. The CV for the baseline (base) and during vocal production (vocMtr) is plotted for each class of neurons. Statistical significance is noted accordingly throughout the figure.

Similar articles

See all similar articles

Cited by 18 articles

See all "Cited by" articles

Publication types

LinkOut - more resources

Feedback