Faculty Bibliography

The relation between the vocal capacities of animals and those of humans is a long-standing topic of interest for scientists, philosophers, and lay people alike. While similar neural and physiological substrates underlie the production of vocal signals in humans and animals, the most celebrated and prototypical aspects of language are cognitive phenomena that go far beyond speech sensorimotor processes. These include a subset of features that have begun to be systematically investigated in nonhuman animals, namely: (i) the presence of statistical laws, (ii) hierarchical syntactic rules, and (iii) the capacity for meaning and reference. Here we review recent progress describing and quantifying language-like structure in animal vocalizations. We highlight agreement and disagreement about the similarities that may exist between human language and animal vocal repertoires, with an eye toward what these phenomena may reveal about the evolution of language and its neural control.

Vocalizations are ancient behaviours that require the complex coordination of breath and display. Understanding how laryngeal anatomy shapes vocalization provides insights into this diversity, its mechanisms and their evolution. Rodents are ideal for exploring this variation because of their diverse mechanisms and vocal structures. Here, we describe the laryngeal morphology and sound production mechanisms underlying the vocalizations of Alston's singing mouse (Scotinomys teguina) and compare these results to those of other vocalizing mammals. We reconstructed the three-dimensional laryngeal morphology with micro-computed tomography, recorded laryngeal sound production using high-speed video and investigated frequency control using surgical ablations. We found that singing mice use a whistle mechanism that uniquely relies on the inflation of an enlarged air sac called the ventral pouch. Song frequency can be controlled by pouch volume, airflow and cricothyroid muscle action. Singing mouse laryngeal morphology and vocal mechanism are distinct from those of other Neotomids; singing mice appear to use inflation-mediated whistles for both distant and close exchanges. Inflatable air sacs have evolved repeatedly for sound modulation and filtering. Our results indicate a novel role for these structures in being required to generate sound. Together, our results expand on an emerging story of how biomechanic and morphological variation contributes to vocal diversity.

Speech production entails several processing steps that encode linguistic and articulatory structure, but whether these computations correspond to spatiotemporally discrete patterns of neural activity is unclear. To address this issue, we use electrocorticography to directly measure the brains of neurosurgical participants performing an interactive speech paradigm. We observe a broad range of cortical modulation profiles, and subsequent clustering analyses establish that responses comprised distinct classes associated with sensory perception, planning, motor execution, and task-related suppression. These activity classes are also localized to separate neural substrates, indicating their status as specialized networks. We then parse dynamics in the planning and motor networks using unsupervised dimensionality reduction, which reveals subnetworks that are sequentially active throughout preparation and articulation. These results therefore support and extend a localizationist model of speech production where cortical activity "flows" within and across discrete pathways during language use.

Alston's singing mice (Scotinomys teguina) are highly vocal Central American rodents that produce structured "songs" (duration: 5-10 s),¹

High-density microelectrode arrays have opened new possibilities for systems neuroscience, but brain motion relative to the array poses challenges for downstream analyses. We introduce DREDge (Decentralized Registration of Electrophysiology Data), a robust algorithm for the registration of noisy, nonstationary extracellular electrophysiology recordings. In addition to estimating motion from action potential data, DREDge enables automated, high-temporal-resolution motion tracking in local field potential data. In human intraoperative recordings, DREDge's local field potential-based tracking reliably recovered evoked potentials and single-unit spike sorting. In recordings of deep probe insertions in nonhuman primates, DREDge tracked motion across centimeters of tissue and several brain regions while mapping single-unit electrophysiological features. DREDge reliably improved motion correction in acute mouse recordings, especially in those made with a recent ultrahigh-density probe. Applying DREDge to recordings from chronic implantations in mice yielded stable motion tracking despite changes in neural activity between experimental sessions. These advances enable automated, scalable registration of electrophysiological data across species, probes and drift types, providing a foundation for downstream analyses of these rich datasets.

Cortical networks for the production of spoken language in humans are organized by phonetic features^1,2, such as articulatory parameters^3,4 and vocal pitch^5,6. Previous research has failed to find an equivalent forebrain representation in other species^7-11. To investigate whether this functional organization is unique to humans, here we performed population recordings in the vocal production circuitry of the budgerigar (Melopsittacus undulatus), a small parrot that can generate flexible vocal output^12-15, including mimicked speech sounds¹⁶. Using high-density silicon probes¹⁷, we measured the song-related activity of a forebrain region, the central nucleus of the anterior arcopallium (AAC), which directly projects to brainstem phonatory motor neurons^18-20. We found that AAC neurons form a functional vocal motor map that reflects the spectral properties of ongoing vocalizations. We did not observe this organizing principle in the corresponding forebrain circuitry of the zebra finch, a songbird capable of more limited vocal learning²¹. We further demonstrated that the AAC represents the production of distinct vocal features (for example, harmonic structure and broadband energy). Furthermore, we discovered an orderly representation of vocal pitch at the population level, with single neurons systematically selective for different frequency values. Taken together, we have uncovered a functional representation in a vertebrate brain that displays unprecedented commonalities with speech-related motor cortices in humans. This work therefore establishes the parrot as an important animal model for investigating speech motor control and for developing therapeutic solutions for addressing a range of communication disorders^22,23.

Inhibitory interneurons are highly heterogeneous circuit elements often characterized by cell biological properties, but how these factors relate to specific roles underlying complex behavior remains poorly understood. Using chronic silicon probe recordings, we demonstrate that distinct interneuron groups perform different inhibitory roles within HVC, a song production circuit in the zebra finch forebrain. To link these functional subtypes to molecular identity, we performed two-photon targeted electrophysiological recordings of HVC interneurons followed by post hoc immunohistochemistry of subtype-specific markers. We find that parvalbumin-expressing interneurons are highly modulated by sensory input and likely mediate auditory gating, whereas a more heterogeneous set of somatostatin-expressing interneurons can strongly regulate activity based on arousal. Using this strategy, we uncover important cell-type-specific network functions in the context of an ethologically relevant motor skill.

Turn-taking is a central feature of conversation across languages and cultures.¹

Neocortical activity is thought to mediate voluntary control over vocal production, but the underlying neural mechanisms remain unclear. In a highly vocal rodent, the male Alston's singing mouse, we investigate neural dynamics in the orofacial motor cortex (OMC), a structure critical for vocal behavior. We first describe neural activity that is modulated by component notes (~100 ms), probably representing sensory feedback. At longer timescales, however, OMC neurons exhibit diverse and often persistent premotor firing patterns that stretch or compress with song duration (~10 s). Using computational modeling, we demonstrate that such temporal scaling, acting through downstream motor production circuits, can enable vocal flexibility. These results provide a framework for studying hierarchical control circuits, a common design principle across many natural and artificial systems.

Bilaterally organized brain regions are often simultaneously active in both humans¹