Faces are among the most relevant visual stimuli for humans as they communicate key aspects essential for social interactions. In psychology, the mechanisms of face perception are traditionally investigated by showing individuals photographs of faces and measuring their behavioral and neurophysiological responses. Recent studies have replaced photographs with virtual faces generated using digital graphics programs. However, it has been suggested that these stimuli might be evaluated differently and elaborated less efficiently than real faces. Some researchers have hypothesized that these differences might be related either to the realism, in terms of render quality, of the virtual faces or to individuals’ expertise with virtual face, in terms of frequency of exposure to these stimuli. However, results are limited and inconsistent and it remains unclear whether virtual faces represent suitable alternatives to real faces in experimental studies. This doctoral dissertation aimed to provide a deeper insight into the role of realism and expertise on virtual faces' evaluation (Chapter 1 and Chapter 2) and recognition (Chapter 3). Chapter 1 described a literature review conducted to establish the link between virtual faces and the uncanny valley effect, according to which highly realistic artificial entities evoke an adverse reaction in the viewer. The results confirmed that virtual faces were judged eerier and less familiar than real faces and that these differences mainly resulted from observers’ perceived inconsistencies in the level of realism of virtual faces. The review further revealed that, so far, nearly no studies considered participants’ expertise with digital characters when analyzing observers’ evaluations of virtual faces. Chapter 2 presented the results of a survey that collected individuals’ judgments of real and virtual faces. Results indicated that virtual faces were judged less reassuring, likable, attractive, trustworthy, and memorable than real faces. The survey also showed that a lower level of realism of virtual faces negatively affected these judgments. This result was especially true for participants being less frequently exposed to digital characters. Chapter 3 presented data on the accuracy with which virtual and real faces were recognized in a memory task. The results showed that virtual faces were recognized less accurately than real faces and that this effect was greater for less realistic virtual faces. This difference was found in individuals with little experience with digital characters and not in experts. Overall, this thesis confirmed that, to date, virtual faces are not an adequate alternative to real faces and that researchers should be cautious when using these stimuli in experiments involving face processing. However, based on the results on realism, it can be hypothesized that advances in digital graphics may overcome this issue by reducing the discrepancies between virtual and real faces. Finally, based on the results on expertise, it is also plausible that virtual faces’ evaluations and elaboration will benefit from the increasing exposure to digital entities in our everyday life.
I volti sono tra gli stimoli visivi più rilevanti per l’uomo poiché comunicano aspetti chiave per le interazioni sociali. In psicologia, i meccanismi di percezione dei volti vengono tradizionalmente studiati mostrando ai partecipanti fotografie di volti e registrando le loro risposte comportamentali e neurofisiologiche. Recentemente, in alcuni studi le fotografie sono state sostituite da volti virtuali creati attraverso programmi di grafica digitale. Tuttavia, è stato suggerito che questi stimoli potrebbero essere valutati ed elaborati diversamente dai volti reali. Alcuni studi hanno ipotizzato che queste differenze potrebbero essere legate sia al grado di realismo, definito dalla qualità grafica, con cui vengono creati i volti virtuali, sia all'esperienza degli individui con questi stimoli, in termini di frequenza di esposizione. I risultati sono limitati e inconsistenti e non è tuttora chiaro qualora l’utilizzo dei volti virtuali in studi sperimentali rappresenti una valida alternativa ai volti reali. Questa tesi di dottorato mira a fornire una visione più approfondita del ruolo del realismo e dell’esperienza sulla valutazione (Capitolo 1 e Capitolo 2) e il riconoscimento (Capitolo 3) dei volti virtuali. Nel Capitolo 1 viene descritta una revisione della letteratura condotta per stabilire il legame tra i volti virtuali e l’effetto uncanny valley, secondo cui le entità artificiali antropomorfe altamente realistiche scaturiscono delle sensazioni di spiacevolezza e repulsione in chi le osserva. I risultati hanno confermato che i volti virtuali vengono giudicati più inquietanti e meno familiari dei volti reali e che ciò potrebbe derivare da delle discrepanze nel livello di realismo dei volti virtuali percepite dagli osservatori. La revisione ha inoltre sottolineato che, ad oggi, quasi nessun studio ha considerato l’esperienza dei partecipanti con i personaggi virtuali come possibile interpretazione dei risultati. Nel Capitolo 2 sono presentati i risultati di un’indagine condotta sulla valutazione di volti reali e virtuali che indicano che questi ultimi siano giudicati meno rassicuranti, piacevoli, attraenti, affidabili e memorabili dei volti reali. I dati dell’indagine hanno inoltre evidenziato che un livello di realismo più basso dei volti virtuali influisce negativamente su questi giudizi e che ciò è vero specialmente per i partecipanti con una minor frequenza d’esposizione ai personaggi virtuali. Nel Capitolo 3 sono esposti i dati relativi all’accuratezza con cui vengono riconosciuti volti virtuali e reali in un compito di memoria. I risultati hanno evidenziato che i volti virtuali sono riconosciuti meno accuratamente rispetto ai volti reali e che questo effetto è maggiore per i volti virtuali meno realistici. Questa differenza è stata riscontrata negli individui con poca esperienza con i personaggi digitali e non negli esperti. Complessivamente questa tesi conferma che, ad oggi, i volti virtuali non rappresentino un'alternativa adeguata ai volti reali e che i ricercatori dovrebbero essere cauti nell’utilizzare questi stimoli in esperimenti inerenti alla percezione dei volti. Tuttavia, alla luce dei risultati sul realismo, si può ipotizzare che i progressi della grafica digitale possano risolvere questo problema riducendo le discrepanze tra volti virtuali e reali. Infine, sul piano dell’esperienza è inoltre ipotizzabile che l’elaborazione dei volti virtuali beneficerà della crescente esposizione ad entità digitali nella nostra vita quotidiana.
(2022). Elaboration and evaluation of virtual faces. (Tesi di dottorato, Università degli Studi di Milano-Bicocca, 2022).
Elaboration and evaluation of virtual faces
DI NATALE, ANNA FLAVIA
2022
Abstract
Faces are among the most relevant visual stimuli for humans as they communicate key aspects essential for social interactions. In psychology, the mechanisms of face perception are traditionally investigated by showing individuals photographs of faces and measuring their behavioral and neurophysiological responses. Recent studies have replaced photographs with virtual faces generated using digital graphics programs. However, it has been suggested that these stimuli might be evaluated differently and elaborated less efficiently than real faces. Some researchers have hypothesized that these differences might be related either to the realism, in terms of render quality, of the virtual faces or to individuals’ expertise with virtual face, in terms of frequency of exposure to these stimuli. However, results are limited and inconsistent and it remains unclear whether virtual faces represent suitable alternatives to real faces in experimental studies. This doctoral dissertation aimed to provide a deeper insight into the role of realism and expertise on virtual faces' evaluation (Chapter 1 and Chapter 2) and recognition (Chapter 3). Chapter 1 described a literature review conducted to establish the link between virtual faces and the uncanny valley effect, according to which highly realistic artificial entities evoke an adverse reaction in the viewer. The results confirmed that virtual faces were judged eerier and less familiar than real faces and that these differences mainly resulted from observers’ perceived inconsistencies in the level of realism of virtual faces. The review further revealed that, so far, nearly no studies considered participants’ expertise with digital characters when analyzing observers’ evaluations of virtual faces. Chapter 2 presented the results of a survey that collected individuals’ judgments of real and virtual faces. Results indicated that virtual faces were judged less reassuring, likable, attractive, trustworthy, and memorable than real faces. The survey also showed that a lower level of realism of virtual faces negatively affected these judgments. This result was especially true for participants being less frequently exposed to digital characters. Chapter 3 presented data on the accuracy with which virtual and real faces were recognized in a memory task. The results showed that virtual faces were recognized less accurately than real faces and that this effect was greater for less realistic virtual faces. This difference was found in individuals with little experience with digital characters and not in experts. Overall, this thesis confirmed that, to date, virtual faces are not an adequate alternative to real faces and that researchers should be cautious when using these stimuli in experiments involving face processing. However, based on the results on realism, it can be hypothesized that advances in digital graphics may overcome this issue by reducing the discrepancies between virtual and real faces. Finally, based on the results on expertise, it is also plausible that virtual faces’ evaluations and elaboration will benefit from the increasing exposure to digital entities in our everyday life.File | Dimensione | Formato | |
---|---|---|---|
phd_unimib_749637.pdf
embargo fino al 19/05/2025
Descrizione: Documento tesi
Tipologia di allegato:
Doctoral thesis
Dimensione
2.75 MB
Formato
Adobe PDF
|
2.75 MB | Adobe PDF | Visualizza/Apri Richiedi una copia |
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.