Search bioRxiv⌕ Search

Biology subjects

Pell, M. D.

Publications and source records attributed to Pell, M. D..

3 recordsLinked to original sources

How do we align in good conversation? Investigating the link between interaction quality and multimodal interpersonal coordination

Conversations can feel effortlessly engaging or, conversely, difficult and unrewarding. Multiple factors contribute to the experienced quality and outcomes of a conversation, among them how interlocutors align with each other. The present study investigated speech-to-speech, brain-to-speech, and brain-to-brain coordination as markers of interpersonal alignment, examining their relationship with jointly perceived interaction quality and mutual affinity between conversational partners. Pairs of previously unacquainted participants (dyads) engaged in multiple short, free-form conversations on topics of varying interest while their vocal and neural activity were simultaneously recorded in a dual-EEG ("hyperscanning") setup. We analyzed interlocutors prosodic adaptation, neural speech tracking, and neural coordination during each conversation. At the speech-to-speech level, our findings reveal that partners with more positive mutual impressions became more similar in their volume and voice quality over the course of the experiment session, reflecting greater prosodic convergence. At the brain-to-speech level, we found no reliable effect of interaction quality on neural tracking of unfolding speech within any individual region, although topographical differences suggested relative modulation across scalp sites. Finally, at the brain-to-brain level, our findings show that higher perceived interaction quality enhanced inter-brain relationships across frequency bands (alpha and theta) and temporal dependencies (concurrent/near-instantaneous and recurrent/listener-lagging), with the strongest effects observed for concurrent alpha-band coupling. These findings suggest that distinct coordination processes are involved in how interlocutors experience an interaction and how they establish relational affinity, casting new light into the mechanisms that make a conversation worthwhile.

neuroscience↗

Human and AI voice identities evoke shared neural signatures during speaker recognition across changes in speech content and prosody

Both biologically-produced human voices and algorithmically-generated AI speech manifest speaker identity. Critically, prosodic variations modulate the acoustic dimensions (e.g., fundamental frequency) that also shape individual speaker identity representations. So far it remains unclear whether listeners process speaker identities in human and AI voices through neurologically equivalent mechanisms, nor how prosodic cues might influence these cognitive processes. We examined event-related potentials during old/new speaker discrimination after name-based identity learning, and further analyzed correctly recognized old speakers comparing trials where prosody matched vs. mismatched between learning and testing. For old/new discrimination, multivariate pattern analysis (MVPA) revealed three significant late windows (662-1498 ms) with Pz as the primary contributor for AI voices, yet none for human voices. Univariate analyses revealed that human voices showed earlier widespread discrimination (N250: 200-280 ms), while both voice types converged on Pz as the strongest contributor based on effect size rankings for late old/new effects (400-800 ms). These old/new effects emerged across completely different speech content between learning and testing, addressing a gap in prior literature. For speaker-specific prosodic expectation effects in the 500-900 ms window, unexpected prosody elicited late positivity for human voices compared to the prosody used during learning, whereas AI voices elicited late negativity. The late positivity resembles P600 components observed for communicative style expectancy violations, while the late negativity likely reflects effortful reprocessing of prosodic violations within atypical synthetic signals, analogous to accented speech processing. Our study advances understanding of voice identity in cognitive neuroscience and offers implications for AI voices in human-computer interaction. [Word count: 250]

neuroscience↗

Speak or shout? Nonverbal vocalizations ensure rapid detection of emotions in vocal communication

Human vocal expressions of emotion can be expressed nonverbally, through vocalizations such as shouts or laughter, or speakers can embed emotional meanings in language by modifying their tone of voice ("prosody"). Is there evidence that nonverbal expressions promote "better" (i.e., more accurate, faster) recognition of emotions than speech, and what is the impact of language experience? Our study investigated these questions using a cross-cultural gating paradigm, in which Chinese and Arab listeners (n=25/group) judged the emotion communicated by acoustic events that varied in duration (200 milliseconds to the full expression) and form (vocalizations or prosody expressed in listeners native, second or foreign language). Accuracy was higher for vocalizations overall, but listeners were markedly more efficient to form stable categorical representations of the speakers emotion from vocalizations (M = 417ms) than native prosody (M = 765ms). Language experience enhanced recognition of emotional prosody expressed by native/ingroup speakers for some listeners (Chinese) but not all (Arab), emphasizing the dynamic interplay of socio-cultural factors and stimulus quality on prosody recognition which occurs over a more sustained time window. Our data show that vocalizations are functionally suited to build robust, rapid impressions of a speakers emotion state unconstrained by the listeners linguistic cultural background.

animal behavior and cognition↗