Behavioral mimicry is commonly assumed to enhance interpersonal liking, yet reported effect sizes vary considerably across experimental studies. Although mimicry manipulations in being-mimicked paradigms share some common features across studies, control conditions differ more substantially in whether confederates remain motorically inactive (no movement; NM) or display natural, noncontingent movements (responsiveness; RES). Such differences may systematically influence participants’ evaluations and, consequently, the estimated magnitude of mimicry-liking effects. To examine this possibility, we conducted a reconstructed meta-analysis, defined here as a synthesis in which published aggregate data from eligible studies were extracted, harmonized, and reanalyzed within a common statistical framework. Major bibliographic databases and supplementary sources were searched, and studies were screened using predefined criteria. Study-level means, standard deviations, and sample sizes were transformed to a common 1–7 liking scale. We then coded confederate motor behavior in control conditions and tested it as a study-level moderator of mimicry-liking effect sizes using weighted meta-analytic models. Across 26 reconstructed contrasts, mimicry showed a robust positive effect on liking, estimated d=0.75, 95% CI [0.56, 0.94]. Importantly, liking in mimicry conditions was highly consistent across studies and implementations. In contrast, evaluations in control conditions varied substantially depending on their design. NM controls produced the lowest baseline liking and the largest mimicry-control differences, estimated d=0.98, 95% CI [0.75, 1.20], whereas RES controls yielded smaller and more stable effects, estimated d=0.36, 95% CI [0.20, 0.52]. These findings suggest that mimicry is associated with higher liking, but that the observed effect magnitude is partly shaped by comparison-condition design. Controls that suppress expressive behavior may artificially lower baseline liking and inflate apparent mimicry effects, underscoring the need for conceptually valid and standardized control conditions in research on mimicry and interpersonal liking.
This article examines designers’ practices related to the use of artificial intelligence (AI)-based tools across different stages of the design process, interpreting them, among other perspectives, through the lens of Actor-Network Theory (ANT). The analysis is grounded in an understanding of design as a complex, staged, and non-linear process that integrates analytical reasoning, intuition, and embodied practical knowledge. The study is based on 14 in-depth individual interviews conducted in 2025 with practising designers.
The analysis focuses on three dimensions of AI’s impact: process acceleration, creativity, and designer responsibility, relating them to the stages of research, conceptualisation, design, and prototyping. The findings indicate that designers primarily perceive AI as a tool that accelerates work, automates repetitive and technical tasks, and supports data analysis and the generation of alternative design solutions. At the same time, AI’s creativity is viewed as secondary and dependent on existing data, whereas the value of design outcomes is understood to derive from human interpretation, intuition, and contextual knowledge. Responsibility for design decisions and their consequences remains with the designer, whose role is increasingly shifting toward that of a curator and critical decision-maker.
Interpreting the findings through the lens of ANT reveals that AI, as a non-human actor, not only reshapes the relationship between designers and technology but also activates strategies aimed at protecting designers’ autonomy, alongside narratives emphasising the importance of creative authenticity. From this perspective, design practices are accompanied by processes of rationalisation and resistance in response to AI’s expanding role in the design process.
This article addresses the significant scarcity of forensic tools for generating voice samples for identification parades by proposing a novel method for voice reconstruction without a reference recording. The presented solution serves as an auditory analogue to facial composites, utilizing a hybrid architecture based on machine learning and speech synthesis (XTTS). The methodology combines an iterative algorithm for selecting a base voice candidate with a specialized neural network module that allows for the modification of interpretable acoustic parameters via latent space manipulation. Technical validation confirmed the system’s effectiveness in navigating the voice space and accurately translating physical parameters into vector representations. By shifting the identification burden from error-prone verbal descriptions to direct auditory perception, the system minimizes the verbal overshadowing effect. Consequently, the proposed prototype offers promising practical implications for forensic science, providing a technological foundation for law enforcement to conduct accurate voice lineups based solely on witness memory.
