Pyry Heikkinen

Pyry Heikkinen

Auto-generated profile, based on publicly available information

Pyry Heikkinen

Cognitive Brain Research Unit, Centre of Excellence in Music, Mind, Body and Brain, Department of Psychology, Faculty of Medicine, University of Helsinki, Helsinki, Finland

Research Focus & Metrics

Academic Career Length

4Years (2023-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
8 March 2023 in Neurobiology of Language
Verified
Journal Article
Open access
PDF file
LTLili Tian
Lili Tian
HCHongjun Chen
Hongjun Chen
PHPyry Petteri Heikkinen
Pyry Petteri Heikkinen
WLWenya Liu
Wenya Liu
TPTiina Parviainen
Tiina Parviainen
The involvement of the motor cortex in language understanding has been intensively discussed in the framework of embodied cognition. Although some studies have provided evidence for the involvement of the motor cortex in different receptive language tasks, the role that it plays in language perception and understanding is still unclear. In the present study, we explored the degree of involvement of language and motor areas in a visually presented sentence comprehension task, modulated by language proficiency (L1: native language, L2: second language) and linguistic abstractness (literal, metaphorical, and abstract). Magnetoencephalography data were recorded from 26 late Chinese learners of English. A cluster-based permutation F test was performed on the amplitude of the source waveform for each motor and language region of interest (ROI). Results showed a significant effect of language proficiency in both language and motor ROIs, manifested as overall greater involvement of language ROIs (short insular gyri and planum polare of the superior temporal gyrus) in the L1 than the L2 during 300–500 ms, and overall greater involvement of motor ROI (central sulcus) in the L2 than the L1 during 600–800 ms. We interpreted the over-recruitment of the motor area in the L2 as a higher demand for cognitive resources to compensate for the inadequate engagement of the language network. In general, our results indicate a compensatory role of the motor cortex in L2 understanding.

22 October 2026 in PsyArXiv
Verified
Preprint
Open access
VHVille Johannes Harjunen
Ville Johannes Harjunen
LLLeevi Lintunen
Leevi Lintunen
SHSami Hyttinen
Sami Hyttinen
PGPatrick Ghaemmaghami
Patrick Ghaemmaghami
SLSanteri Lepistö
Santeri Lepistö
EKEmmi Koskinen
Emmi Koskinen
Little is known about how Big Five personality traits manifest in specific conversational behaviors or whether these associations depend on the communication medium. We examined 154 individuals interacting with strangers face-to-face or via video. Using manual annotations, computer vision, and LLMs, we examined actor and partner personality associations with turn-taking dynamics, engagement signalling, and topic development. Higher actor and partner extraversion predicted faster turn-taking and pronounced engagement signalling, whereas actor openness predicted slower turn-taking and reduced engagement signalling. In actors, conscientiousness predicted nodding and in partners it predicted broader topic exploration. However, we found no evidence that these associations differed between face-to-face and video-mediated interaction. Personality thus shapes conversational behavior and these interactional fingerprints appear to survive technological transmission.

30 March 2026 in SocArXiv
Verified
Preprint
Open access
MSMarkus Söderman
Markus Söderman
SLSanteri Lepistö
Santeri Lepistö
EKEmmi Koskinen
Emmi Koskinen
PHPyry Heikkinen
Pyry Heikkinen
ASAino Sinisalo
Aino Sinisalo
VHVille Johannes Harjunen
Ville Johannes Harjunen
Backchanneling is a core interactional practice through which listeners display attention, engagement, and affiliation during a conversation. While backchannels are used among most language communities, prior research suggests systematic variation across language groups in frequency, type and timing of these actions. However, following the rapid expansion of video-mediated interaction, it has become evident that video-calling constricts provision of well-timed listener feedback. It is therefore increasingly important to understand how sociocultural differences in backchanneling are reshaped under such technological constraints. In the present study, we examined backchanneling behavior during first encounters among 154 Finnish-speaking and Swedish-speaking Finns, representing relatively low- and high-engagement cultures, respectively. Dyads were randomly assigned to either face-to-face or video-call. We analyzed multiple forms of backchanneling (lexical and non-lexical vocal responses, and nods) to capture their distinct interactional functions and modality-specific affordances. Consistent with prior work, Swedish speakers exhibited higher rates of backchanneling than Finnish speakers in face-to-face interaction. However, these sociocultural differences were substantially reduced during video-calling. Especially non-lexical backchannels and nodding showed clear language-related differences in face-to-face settings that attenuated in video calls. Time-course analyses further revealed that the technologically contingent sociocultural differences in backchanneling were relatively stable across conversation, especially when it comes to nodding. Together, the findings suggest that video-calling constrains interactional timing and multimodal signaling in ways that reduce sociocultural variability in backchanneling. This suggests that the expansion of video-calling may place high- and low-engagement cultures in unequal positions, disproportionately constraining the more expansive backchanneling practices characteristic of high-engagement interaction style.

26 October 2026 in The Journal of Neuroscience
Verified
Journal Article
Closed access
PHPyry Heikkinen
Pyry Heikkinen
SLSanteri Lepist
Santeri Lepist
PWPatrik Wikman
Patrik Wikman
EKEmmi Koskinen
Emmi Koskinen
APAnssi Perkyl
Anssi Perkyl
VHVille Johannes Harjunen
Ville Johannes Harjunen
Engagement in conversation often requires continuous monitoring of relevant speech in a continuously evolving auditory environment. Emotional salience of semantic content may influence this process, but its role in attentional tracking of continuous speech remains unclear. We address this question using an audiovisual cocktail-party paradigm combined with source-localized magnetoencephalography and encodingdecoding analyses. Participants (18 females, 14 males) selectively attended to one dialogue while ignoring a concurrent distractor dialogue, with both emotional salience and semantic coherence of the conversations manipulated independently. Decoding analyses revealed that negatively valenced, high-arousal, and socially relevant semantic content strengthened neural speech tracking, but only for attended dialogues. This effect depended on semantic structure, emerging earlier in semantically coherent than incoherent dialogues. Encoding analyses further showed latency- and region-specific effects across temporal, parietal, and frontal cortical regions. Emotional salience and coherence jointly modulated midlatency (100250 ms) responses, whereas late-latency (270450 ms) responses were driven primarily by emotional salience. These findings demonstrate that emotional meaning enhances neural prioritization of attended speech and semantic coherence shapes the temporal dynamics of this enhancement. Together, these results indicate context-sensitive emotional modulation of auditorylanguage processing during selective listening of continuous speech.

16 June 2026 in PsyArXiv
Verified
Preprint
Open access
SLSanteri Lepistö
Santeri Lepistö
PHPyry Heikkinen
Pyry Heikkinen
DTDiana Tamir
Diana Tamir
EKEmmi Koskinen
Emmi Koskinen
MSMarkus Söderman
Markus Söderman
VHVille Johannes Harjunen
Ville Johannes Harjunen
Conversation partners signal their engagement in multiple ways, including eye gaze, nodding, and facial expressions. While disruptions to this engagement—whether due to listeners’ divided attention or constraints of the communication medium—are associated with changes in speakers’ stress levels, it remains unclear how specific expressions of listener (dis)engagement drive fluctuations in speakers’ emotional states on a moment-to-moment basis. Here, we analyzed naturalistic conversations in face-to-face and video-mediated settings to test how changes in listeners’ attentional engagement translate into observable engagement behaviors, and how these behaviors, in turn, influence speakers’ emotions. A sample of 160 participants (data collected in 2023–2024) engaged in a dyadic emotional self-disclosure task either face-to-face or via video call, with listeners interacting either with full attention or in the presence of an attentional distractor. During the interactions, we tracked listeners’ engagement behaviors (nodding, gaze, smiling) and changes in speakers’ emotional state (autonomic arousal). The results showed that attentionally engaged listeners displayed more engagement behaviors, which in turn regulated speakers’ emotional responses across different timescales. At the level of the whole encounter, listeners’ active nodding during attentional disengagement reduced speakers’ autonomic arousal. At the moment-to-moment level, listeners’ nodding, gaze diversion, and smiling all transiently increased speakers’ arousal. The effect of gaze diversion was weaker in video-call interactions, whereas the effect of smiling was stronger under full attention than under attentional distraction. These findings provide new insight into listener–speaker emotional dynamics and how they are disrupted in online interaction and under divided attention.

7 November 2025 in bioRxiv
Verified
Preprint
Open access
PHP. Heikkinen
P. Heikkinen
SLS. Lepistö
S. Lepistö
PWP. Wikman
P. Wikman
EKE. Koskinen
E. Koskinen
APA. Peräkylä
A. Peräkylä
VHV. J. Harjunen
V. J. Harjunen
Engagement in interaction arises through the continuous monitoring of conversation content, allowing individuals to detect and act upon information that is personally relevant or socially significant. Emotional salience of speech content is a key factor in guiding this attentional engagement. By combining an audiovisual “Cocktail-party” paradigm with source-localized magnetoencephalography and encoding–decoding models, we examined whether emotionally salient content continuously enhances attentional tracking of conversations. Decoding analyses showed that emotional salience strengthened neural speech tracking across the conversational arc, but only when dialogue was attended to. The emotional enhancement was also sensitive to the dialogue’s semantic structure, peaking earlier in semantically coherent than incoherent conversations. Encoding analyses revealed latency-specific effects across temporal, parietal and frontal regions. Emotional salience and semantic structure co-modulated mid-latency sensory encoding, whereas later semantic integration was driven by emotionality alone. The observed results support the notion of adaptive tuning of auditory–language networks to socially and motivationally significant semantic cues. Rather than acting as a simple amplifier of semantic processing, emotional salience of dialogue topics shapes neural speech processing at multiple stages. The exact modulations depend on the semantic structure of the dialogue, balancing the needs between efficiency and adaptability that together support the ability to engage optimally in real-world interactions. Significance statement: The emotionality of a conversation is a key factor in how people allocate attention during interaction. Using a multi-speaker “cocktail-party” paradigm with magnetoencephalography, we show that emotionally salient content and semantic coherence influence the brain’s tracking of continuous dialogues in the presence of competing speech. Emotionally salient content enhances neural tracking only for attended speech, and this enhancement appears earlier when the conversation consists of semantically interrelated turns and is easier to follow. This is supported by modulations in neural activity at distinct processing stages across a distributed cortical network. These findings indicate that emotionality functions as a relevance signal, sustaining attentional engagement in conversation by affecting predictive processing and modulating activations in a large-scale cortical system linked to speech comprehension.