Search

Virtual Reality (VR) is widely used in various sectors, however problems of cybersickness (motion sickness symptoms induced by VR) and perceptual aftereffects remain barriers for uptake. These negative problems may be caused by conflicts between sensory modalities which signal self-motion, such as vision, proprioception, and the vestibular system (Gallagher & Ferrè, 2018; Rebenitsch & Owen, 2016). These modalities are also implicated in the perception of gravity (Jörges & López-Moliner, 2017; Lacquaniti et al., 2015). Accordingly, it may be possible that gravitational conflicts may cause cybersickness and VR-induced aftereffects. We are currently undertaking a study investigating whether reducing visuo-gravitational conflicts in VR can reduce cybersickness and VR-induced aftereffects, as well as improve the VR experience (https://osf.io/6zpbf). Participants will play a VR game set in a Zero-Gravity environment while their bodyweight is supported by an anti-gravity treadmill (Low-G Condition). In this condition, visual, proprioceptive and somatosensory signals indicate that the participant is in a reduced-gravity environment, reducing visuo-gravitational conflicts. In a control condition (Normal-G), bodyweight is not supported by the treadmill, creating conflicts between vision and body-related gravity signals. We predict lower cybersickness, fewer aftereffects, and increased presence and immersion in the experimental condition. In the present study we aim to investigate the stability of responses to VR gravitational conflicts by re-testing individuals from 5-12 days following the initial study. All procedures will remain identical. Analysis of this data will only be conducted if at least two-thirds of the original sample (n=40, 20 per condition) return for the re-test session.

Virtual Reality (VR) has gained significant popularity since the release of commercial Head-Mounted Displays, with applications ranging from recreation and gaming to training and rehabilitation. Despite advancements in VR technology, implications of VR on cognition and perception have not been fully explored. One outstanding issue which may hamper VR uptake and impair the user experience is Cybersickness, symptoms of motion sickness induced by VR exposure. Cybersickness is believed to arise from conflicts between sensory modalities signalling self-motion (Gallagher & Ferrè, 2018; Rebenitsch & Owen, 2016). For example, when viewing a VR rollercoaster, vision signals that the user is moving in a certain direction with a certain acceleration, while the vestibular system signals that the user is stationary. These visuo-vestibular self-motion conflicts may also cause aftereffects following VR exposure, such as altered vestibular perception and reflexes, and disrupted proprioceptive coordination (Di Girolamo & Pic, 2001; Gallagher et al., 2019, 2020; Harm et al., 2008). While visuo-vestibular self-motion conflicts have been given some attention in the literature, far less attention has been given to conflicts between other senses which signal self-motion, for example proprioception and somatosensation. Moreover, the sensory modalities implicated in self-motion perception are also crucial for the perception of gravity (Jörges & López-Moliner, 2017; Lacquaniti et al., 2015), and it is unclear whether conflicts relating to gravity (such as vision signalling an altered orientation or reduced strength of gravity with respect to true gravity) may also cause cybersickness and VR aftereffects. The aim of our research is to investigate whether reducing visuo-gravitational conflicts can reduce cybersickness and VR aftereffects, as well as improve the overall VR experience. Participants will play an off-the-shelf VR application which visually signals that they are in a Zero-Gravity environment. In the experimental condition (Low-G), visuo-gravitational conflicts will be reduced by supporting the participants’ bodyweight on an Anti-gravity Treadmill (Alter-G) such that the participant feels almost weightless. Accordingly, visual, proprioceptive and somatosensory signals will all indicate that the participant is in a reduced-gravity environment, reducing visuo-gravitational conflicts. In the control condition (Normal-G), participants will stand on the treadmill but they will remain at 100% body weight, creating gravitational conflicts between visual and proprioceptive/somatosensory systems. In both conditions, we will take objective (heart rate, blood pressure, Dennison et al., 2016) and subjective (Kennedy et al., 1993; Keshavarz & Hecht, 2011) measures of cybersickness, while VR aftereffects will be assessed through a psychophysical verticality detection task (Arshad et al., 2023). The user experience will be assessed through measuring immersion (the feeling of engagement by the virtual environment, Jennett et al., 2008) and presence (the feeling of being located within the virtual environment, Witmer & Singer, 1994).

This is a pre-registration of a project investigating decision making in unfamiliar face identification. Information for this pre-registration can be found in the uploaded document titled "Pre-reg_FaceDecisions-Exp3_BakerBindemann.docx".

This project aims to investigate decision making in unfamiliar face identification.

This is a pre-registration of the second experiment of a project investigating decision making in unfamiliar face identification

This experiment seeks to establish the test-retest reliability of a new face matching task (VRFMT) that is designed to be performed in Virtual Reality. In this task, observers match photographs of real faces to computer avatars that have been constructed from 3D face scans. We will compare this with two established identity matching tests (GFMT, KFMT) that require comparisons of pairs of face photographs. Each participant will perform the three tasks twice, with an interval of approximately 7 days between testing sessions 1 and 2.

The aim of this experiment is to compare performance of a new face matching task, in which observers match photographs to images of digitally-rendered people (“avatars”) derived from virtual reality (VR) software, with two already established face-matching tasks that are based exclusively on comparisons of face photographs (Glasgow Face Matching Task, GFMT; Burton, White, McNeill, 2010; Kent Face Matching Task, KFMT; Fysh & Bindemann, 2018). Previous findings suggest that the decision of whether two different face images belong to the same person identity is difficult and error prone. The present experiment aims to extend those findings with a new task, in which ID style photos will be matched with face portraits of avatars (2D face images derived from VR technology). The study will test the validity of the VR stimuli for further use in face matching research. Participants will be asked to decide whether a pair of images they see on a computer screen depicts the same person or two different people. They will perform three different matching tasks with three different sets of stimuli – the VR Face Matching Test (VRFMT), the GFMT and the KFMT. Each task will contain match and mismatch trials, as well as attention check trials in which participants will need to respond with a different key to upside down face pairs.

This experiment compares performance of a new face matching task, in which observers match photographs of real faces to 3D avatars in virtual reality (VR) software, with two already established face-matching tasks that are based exclusively on comparisons of face photographs (Glasgow Face Matching Test, GFMT; Burton, White, & McNeill, 2010; Kent Face Matching Test, KFMT; Fysh & Bindemann, 2018). In all three tasks, participants will be asked to decide whether pairs of visual stimuli (two face photos, or face photo-and-avatar) depict the same person or two different people. In a previous experiment in which the matching of static images of avatars was compared with the GFMT and KFMT (osf.io/36x4d), we obtained correlations in accuracy between these three tasks. The aim of this experiment is to replicate these findings.

We have recently assembled a stimulus database of 120 avatars (i.e. digital representations of people) that were constructed from 3D facial scans of actual faces. We plan to use these avatars to study person recognition in virtual reality (i.e. VR). The purpose of this experiment is to establish behavioural characteristics of these avatars, by contrasting identity matching of familiar and unfamiliar faces. To do this, we will recruit observers who are already familiar with a subset of the avatars in our database. Across 80 trials (40 match, 40 mismatch), observers will view an avatar alongside a digital face photograph, and will decide whether these together show the same person (a match) or two different identities (a mismatch). Upon completion of this task, observers will then complete a familiarity check to establish the identities with which they were already familiar.

This study examines whether individual differences in normal and corrected-to-normal visual acuity (i.e., with glasses) affect face processing. Participants will complete standard acuity tests, comprising the AOE and HAL Snellen charts, as well as the Freiberg Vision Test (Bach, 1996). These will be performed with and without glasses to establish both corrected and uncorrected acuity. This will be followed by three face perception tasks, comprising the Kent Face Matching Test (KFMT; see Fysh & Bindemann, 2018), the Cambridge Face Memory test (CFMT+; see Duchaine & Nakayama, 2006), and the Cambridge Face Perception Test (CFPT; see Duchaine, Germine, & Nakayama, 2007). Participants with glasses will perform the face tests twice, with and without acuity correction, to provide a control condition that should reveal clear acuity effects in face matching even if finer individual differences with corrected vision are not found. Viewing distance will be maintained via a chinrest.

This project has investigated the number of unique face identities (“set size”; either 2, 4, 6, 8, 10, 12, 14, 16, 18, 20, 30, or 40) required before facial averages are judged to be depicting the same identity (“perceptual stability”). In these experiments, participants were instructed to judge if the identities of two simultaneously displayed facial averages depicted the same identities or different people (a two-alternative forced-choice (2AFC) comparison task). The aim of the previous experiment (see “Variation in the perceptual stability of cross-identity facial averages”) was to test whether stability varied according to the number of different face categories used to construct the averages. Six conditions were included: two ‘within-category’ and four ‘between-category’. Facial averages were constructed from either one face category (i.e. ‘within-category’; both young Caucasian males and young Caucasian females), two face categories (varying across gender: male or female), four face categories (varying across gender and also age: old or young), six face categories (varying across gender and also race: Caucasian, African or Asian) or twelve face categories (varying across gender, age and race). Previous experiments in this project (see “The perceptual stability of cross-identity facial averages”, “The perceptual stability of cross-identity facial averages between-categories”, and “Replicating the perceptual stability of cross-identity facial averages”) have reported that within-category and between-category averages converge at set size 30. Contrary to this previous work, we found that differences between conditions persisted into the higher set sizes (e.g. set sizes 30 and 40), despite there being little variance present across the average images themselves. This discrepancy could be due to the demands of the task. The previous experiment used a fully randomised design, presenting face pairs from any of the six conditions throughout. The other experiments in this project presented stimuli from each condition separately, using a block design. If participants use the context of recently presented trials to inform their match decisions, within a block design the most similar face pairs would simply be those of the highest set sizes, hence participant responses would converge across conditions. However, in a fully randomized design the most similar face pairs in each ‘block’ would be those from the highest set sizes within low-variability conditions (i.e. within-category conditions), which would lead to differences across conditions. Participants may also have become sensitive to small differences between face averages given the large trial number (N = 864), a large proportion of which would have been match decisions. To test this possibility, we plan to repeat the previous experiment using a block design. At the time of writing, data collection in laboratories is not possible due to the Covid-19 lockdown. Future experiments in this project will therefore be conducted online. To test whether online data collection produces comparable results to lab testing, we will conduct an additional experiment that attempts to directly replicate the previous experiment (“Variation in the perceptual stability of cross-identity facial averages”) using an online format.

Observer's fixations to Regions of Interest (ROIs) for the male and female stimuli across exposure conditions

This provides an additional analysis of the data from both experiments, where increasing and decreasing time pressure are treated as separate conditions.

Research highlights that we are not as skilful in controlling our memory as we may believe. Instead, our everyday intentional recognition judgments are often biased by what we unintentionally recognise in the same context. So far, it has been demonstrated that the unintentional recognition of image distractors can bias the intentional recognition of word targets, in the form of a familiarity (old/new) congruency bias. This bias reflects improved recognition performance for targets when the distractor/context upon which it is present at test is of the same memory status (also old or new). However, this effect has not yet been explored using face stimuli, despite faces varying in pre-existing familiarity and often being encountered in different familiar or unfamiliar contexts in everyday life. Furthermore, the distractor stimuli used in past literature have often been limited to simple drawings. Past designs have also typically relied on the use of working memory load or divided-attention tasks, or healthy aging to magnify distractibility, which is arguably not ecological valid nor generalisable. Consequently, this research investigated whether distractor-induced congruency biases found for words also apply to faces, using a new database of up-to-date face stimuli and without secondary manipulations of distractibility. I also attempted to replicate these results in an alternative sample and compared effects between target types (words vs faces). Results show novel evidence for the idea that faces are also biased by distracting stimuli in the same manner that has been found in relation to words. In turn, providing evidence for specific cognitive theories (e.g. Perceptual load theory) while questioning others (face processing modularity). Lastly, the study also provides future direction for neurocognitive research to answer questions regarding the underlying mechanisms of distractor bias, based on past research findings of dissociating event-related potentials (ERPs) in relation to unintentional and intentional recognition.

Most research on eyewitness memory has focused on single-perpetrator crimes. However, crimes to which eyewitnesses may bear testimony are often committed by groups of perpetrators. A consequence of researching only single-perpetrator crimes is that we know very little about how set size (i.e., the number of faces) at encoding impacts recognition performance. We do not know much more about this question in the face recognition literature either but the small extant literature does appear to converge on one conclusion, namely that recognition performance is worse for larger set sizes. In the case of eyewitness memory, the presence of multiple perpetrators poses an additional unique question: Eyewitnesses not only need to identify perpetrators, but also need to testify to the perpetrators' actions. Few researchers have investigated this second aspect. In this chapter, we review literature in the areas of face recognition and eyewitness memory to shed light on these questions, and present two laboratory studies that test the effects of set size on face and person recognition. Results show that recognition performance decreases as a function of set size, but that this is differentially true for faces and roles, and is in fact dramatically reduced when faces and roles are paired. There are serious applied implications for this latter finding in particular.