I spy with *your* little eye: investigating the remarkable ability to "borrow" the eyes of others.

Background and Significance

Have you ever replayed in your mind, over and over, how your embarrassing mishap must have looked to other people? Does your behaviour change when you’re being watched? Are you good at guessing how others will react in certain situations?

These are all examples of people’s remarkable ability to take someone else’s perspective. We live in a complex social world, full of other people with thoughts, feelings, and beliefs that may differ from our own. To successfully navigate this environment, we need to be able to understand how the world looks through other people’s eyes, so that we can better appreciate what they might be thinking and feeling. Visual perspective-taking is therefore an important stepping-stone towards more sophisticated social abilities such as empathising and mentalising, and provides a pivotal foundation for the remarkable social interaction capabilities that define us as humans.

Despite this importance, visual perspective-taking abilities can vary significantly between different people. While some expert perspective-takers can fluently shift between their own and others’ viewpoints, others find themselves either locked in their own perspective or overcome by that of others, and these problems often go along with pronounced difficulties in social interaction, as seen in conditions such as schizophrenia and autism spectrum conditions (ASC).

Strikingly, despite massive research interest and important real-world implications, research provides almost no clues about what makes someone a good or bad perspective-taker. Are some people simply less alert to when another perspective is available, such as the presence of another person who has a mind? Or is the difficulty in controlling when to perspective-take and when not to (e.g., ignoring that of someone who is watching you)? And are these specific difficulties related to the individual’s personal traits?

This ambitious research project will answer these questions. In two experiments, we will measure people’s tendency to spontaneously form an image of another person’s perspective when available (Experiment 1), and their ability to voluntarily control these perspectives when necessary (Experiment 2) to reveal how they relate to people’s individual differences in social interaction skills, their ability to form mental images, as well as their schizotypal and autistic traits. By combining our well-established visual perspective-taking task with Prolific’s access to large and varied populations, this project will reveal exactly where perspective-taking impairments manifest, providing the groundwork for developing therapies and interventions to train these abilities and ultimately improve social capability.


Both experiments will utilise our task (Ward, Ganis & Bach, 2018; Ward, Ganis, McDonough & Bach, 2019, 2021) that robustly measures people’s visual perspective-taking ability, in both lab and online. Participants simply judge, as quickly as possible, whether a letter appearing in various rotated orientations on a table is normal or mirror inverted (R vs Я). Generally, the more the letter is rotated away from themselves, the harder the task (classic mental-rotation effect). Good perspective-takers, however, can ‘borrow’ the eyes of other people in the scene: they are better able to judge letters rotated away from themselves if they appear upright to another person (Figure 1a), compared to when rotated further away from this other person (Figure 1b). The size of this difference provides a direct measure of how well a participant can access the visual perspective of another. To try the task for yourself, click here.

Experiment 1 uses this task to test people’s spontaneous access to another person’s perspective, and to determine the characteristics of the other person that activates this ability, such as having a “mind”. We will measure visual perspective-taking of different types of agents (e.g., human, robot, mannequin, dog, or lamp, Figure2), but also ask each participant to rate these agents on how human-like they “look” and seem to be able to “think”. This will allow us to measure (a) people’s overall spontaneous perspective-taking ability, delineate how it is triggered by (b) visual and/or (c) mentalistic features, and (d) how individual differences in social skills, imagery ability and schizotypal and autistic traits shape the sensitivity to these features. Click here to try a demo.

Experiment 2 tests people’s ability to control their perspective. We will instruct participants to perform the task either from their own perspective (ignoring that of the other person), or from the perspective of the other person (ignoring their own). This will allow us to measure (a) people’s overall ability to control visual perspective-taking, (b) how well a given perspective can be taken, (c) how well it can be supressed, and (d) how these relate to individual differences . Click here for demo.

Individual difference measures. After each experiment, we will measure people’s ability to form vivid mental images (as measured by the Vividness of Visual Imagery Questionnaire), their general social ability (Interpersonal Reactivity Index questionnaire), the presence of autistic-like traits (Autism Quotient Questionnaire), and the presence of schizotypal traits (Schizotypy Questionnaire). Together, these measures will allow us to robustly link people’s individual characteristics both to their tendency to spontaneously take others’ perspective in Experiment 1, and their ability to control these perspectives in Experiment 2.

Sample size and costs

In both experiments, power is constrained not by the between-condition comparisons, but the goal of correlating these differences to individual differences , which are typically smaller ( r ~.16-.20 in recent meta-analyses across Psychology). A sample size of 795 (anticipating 20% pre-registered exclusions) in each experiment gives 90% power to detect such effect sizes, with a Bonferroni-adjusted p of .004, to correct for multiple-comparisons in correlational analyses. For each 30-minute experiment, paying £4 each, the Prolific-estimated costs are £4,452.00 (total £8904.00).

Summary and outlook

This pre-registered study combines the large and broad samples available through Prolific with a novel and engaging task that showcases what is possible in online research. It promises important new insights into people’s ability to perspective-take, opening-up new avenues for not only understanding the determinants of this uniquely human ability, but also its improvement in psychological conditions. All findings, materials, data and analyses will be made available on the lab website and open-access depositories (OSF, PsyArxiv). We intend to publish in an open access journal.