Rosa Rubicondior: Jesus & Mary On Toast - Why We See Faces Everywhere.

Tuesday, 29 July 2025

Jesus & Mary On Toast - Why We See Faces Everywhere.

New research details how our brains are drawn to and spot faces everywhere | University of Surrey.

From an early age, the human brain interprets patters and sees faces everywhere. Draw two large black discs on a white sheet of paper and show it to a human baby of just a few weeks old and you will probably make the baby smile, just as they smile when they look at your face. Their brain has perceived the dots as the eyes of a face. This is the phenomenon of human perception known as pareidolia.

Some religious people, in an effort to justify holding irrational beliefs, often attribute these appearances of 'Jesus' or 'Mary', as miracles with some deep religious significance. For example, this anti-Atheist Gotcha, which appeared on Quora a couple of years ago:

Atheists, if there is no God, then why does Jesus' face sometimes appear on toast?

Imagine walking through the woods and briefly believing you're seeing a face in a knot of moss—only to realise seconds later that it’s just a tree. That flicker of recognition is your brain’s face-detection machinery springing into action. In a recent experiment at the University of Surrey, researchers used gaze-cueing techniques to show that when people perceive a face-like arrangement in inanimate objects, their attention is captured even more powerfully than by the averted gaze of a real face. These face-like patterns don’t need eyes, a nose, or even realistic features—just the right suggestion of symmetry and placement is enough for the brain to “see” a face where none exists.

This phenomenon, known as face pareidolia, is a by-product of evolution, not evidence of design. From an evolutionary standpoint, it's safer to wrongly detect a face in a shadow than to miss a predator or an ally. But while this trait once had clear survival value, it now fuels more curious beliefs. Religious individuals, for instance, often interpret pareidolic faces—especially when perceived in burnt toast, water stains, or tree bark—as miraculous signs from divine figures like Jesus or the Virgin Mary. What neuroscience tells us is that these “visions” are false positives produced by a brain primed to find patterns and assign meaning, even when none exists. Evolution has given us a perceptual system that’s fast, but not always accurate—and our cultural context then fills in the rest.

How the human brain interprets so many things as faces is the subject of a recent open access paper in i-Perception by researchers led by Dr Di Fu of the University of Surrey, with colleagues from the Institute of Psychology, Chinese Academy of Sciences.

What information is there on the evolutionary advantage of pareidolia in humans? Pareidolia — the tendency to perceive meaningful patterns (especially faces) in random stimuli — is often regarded as a by-product of evolved cognitive mechanisms in humans. While it's not an adaptation in itself, it emerges from systems that are adaptive. Here's a summary of what’s known and hypothesised about its evolutionary roots and advantages:

Face Detection: A High-Stakes Priority

Humans are exceptionally good at detecting faces. This likely evolved because:

Rapid recognition of allies or threats (friend vs foe) was vital for survival.
Infant-caregiver bonding depends on face recognition (e.g. recognising mother’s face).
Social signalling — we communicate emotions, intentions, and identity through facial expressions.

Pareidolia is thought to stem from the brain's hyperactive face detection system — it errs on the side of caution, preferring false positives (seeing a face that isn’t there) over false negatives (failing to spot a real face, which could be dangerous).

The Brain Is a Pattern-Seeking Machine

The human brain evolved to detect structure in noise — whether in the natural environment, in language, or in social behaviour:

Anticipating predators: Seeing a predator in the bushes when it might not be there is less costly than failing to see it when it is.
Detecting agency: The same mechanisms that underlie pareidolia also underlie agency detection, a proposed evolutionary trait that encourages us to attribute movement or change to an intentional agent — which could be another human or a dangerous animal.

This hyperactive agency detection could help explain not only pareidolia, but also the human tendency toward animism, religion, and conspiracy thinking.

Fast Heuristics Over Costly Accuracy

From an evolutionary perspective, speed often outweighs accuracy:

Pareidolia may reflect a "better safe than sorry" heuristic.
The cost of occasionally being wrong (e.g. seeing a "face" in a rock) is low compared to the cost of missing real dangers or opportunities.

This is similar to what's known as Error Management Theory — where evolution favours cognitive biases that minimise the most costly type of error.

Developmental Evidence

Newborns show a preference for face-like patterns (e.g. two dark dots above a line), suggesting that the brain is hard-wired to attend to facial configurations.
Pareidolia is observed early in infancy, even before the child develops a full understanding of what a face is.

Neuroscientific Backing

Brain imaging shows that pareidolia activates the fusiform face area (FFA) — the same region used for processing real faces.
This suggests that pareidolia is not a separate process, but an artefact of the same machinery used for recognising real faces.

Is It Unique to Humans?

Pareidolia-like responses have been observed in primates, especially when it comes to facial recognition.
For example, rhesus monkeys show prolonged interest in objects with face-like symmetry.

This points to a deep evolutionary origin shared among social primates.

Summary: Evolutionary Advantage of Pareidolia

Face detection bias: Helps with social bonding and threat detection
Agency detection: Avoids missing intentional agents (predators, humans)
Pattern recognition: Supports learning, memory, environmental awareness
Error management bias: Favouring false positives over false negatives
Infant facial preference: Encourages early bonding and learning

Their work is explained in a news item from the University of Surrey:

New research details how our brains are drawn to and spot faces everywhere
If you have ever spotted faces or human-like expressions in everyday objects, you may have experienced the phenomenon of face pareidolia. Now, a new study by the University of Surrey has looked into how this phenomenon grabs our attention, which could be used by advertisers in promoting future products.

The study, published in i-Perception, investigated the differences between our attention being directed by averted gazes – when a subject looks away from another subject’s eyes or face – and when it’s directed by pareidolia – imagined face-like objects.

The researchers conducted four “gaze cueing task” experiments with a total of 54 participants, to measure how our attention is influenced by the direction of another subject’s gaze. They found that participants consistently shifted their attention in response to the appearance of both averted gazes and pareidolia.

However, the underlying mechanisms through which attention is drawn are quite different. While we are primarily drawn to the eye region of averted gazes, we are drawn to pareidolia’s holistic structure of their “faces”, and as a result, experienced a stronger response and attention.

Our research shows that both averted gazes from real faces and perceived faces in objects can direct where we look, but they do so through different pathways. We process real faces through focusing on specific features, like the direction of the eyes. However, with face-like objects, we process their overall structure and where their “eye-like features” are positioned, resulting in a stronger attention response.

Dr Di Fu, lead author
School of Psychology
University of Surrey
Guildford, Surrey, UK

The findings of the study may have implications that go beyond a better understanding of how our brain processes information.

Our findings may have practical implications too, particularly in areas like product advertising. Advertisers could potentially incorporate face-like arrangements with prominent eye-like elements into their designs, increasing consumer attention and leaving a more memorable impression of their products.

Dr Di Fu.

Publication:
Chen, Z., Wen, M., Liu, X., & Fu, D. (2025).
How face-like objects and averted gaze faces orient our attention: The role of global configuration and local features. I-Perception, 16(4). https://doi.org/10.1177/20416695251352129.

Abstract
In real life, people perceive nonexistent faces from face-like objects, called face pareidolia. Face-like objects, similar to averted gazes, can direct the observer's attention. However, the similarities and differences in attentional shifts induced by these two types of stimuli remain underexplored. Through a gaze cueing task, this study compares the cueing effects of face-like objects and averted gaze faces, revealing both commonalities and distinct underlying mechanisms. Our findings demonstrate that while both types of stimuli can elicit attentional shifts, the mechanisms differ: averted gaze faces rely on processing local features like gaze direction, whereas face-like objects leverage their global configuration to enhance attentional shifts by triggered eye-like features. These findings advance the understanding of the processing mechanisms underlying the perception of face-like objects, and how the brain represents facial attributes even when physical facial stimuli are absent. This study provides a valuable theoretical foundation for future investigations into the broader applications of face-like stimuli in human perception and attention.

Introduction
Individuals often perceive nonexistent faces from meaningless and ambiguous objects, a phenomenon known as face pareidolia (Palmer & Clifford, 2020; Takahashi & Watanabe, 2013). The detection of face pareidolia involves both local and global processes, engaging both object perception and higher-order cognitive functions (Akdeniz, 2023; Leadner et al., 2022; Lhotka et al., 2023.1; Wardle et al., 2023.2). Perceiving face-like objects is associated with increased prosocial behavior (Hamza et al., 2022.1), influencing individual attention allocation and diversion (Castello et al., 2014; Henschel et al., 2021; Jakobsen et al., 2023.4; Keys et al., 2021.1). For instance, architecture and paintings incorporating face pareidolia can capture attention (Martinez-Conde et al., 2015; Wang et al., 2022.2). Additionally, products incorporating face pareidolia can guide consumers’ attention to increase their purchasing behaviors (Delbaere et al., 2011; DiSalvo & Gemperle, 2003; Guido et al., 2018; Noble et al., 2023.5; Wodehouse et al., 2018.1). However, the mechanisms by which face-like objects trigger attentional shifts remain unclear, as does the distinction between these shifts and those induced by common averted gaze faces.

The low-level physical properties and high-level social attributions of face-like objects likely contribute to attentional shifts. From a physical perspective, individuals might perceive the orientation of objects based on large-scale variations in luminance or asymmetry. However, few studies have investigated the impact of the objects’ orientation on triggering attentional shifts. From a social perspective, individuals might attribute social meanings to these objects, allocating attention similarly to how they would respond to averted gazes or head orientation. Averted gazes often prompt individuals to look in the direction of the gaze to obtain social information (Frischen et al., 2007; Ishikawa et al., 2021.2; Sweeny & Whitney, 2014.1; Tanaka et al., 2022.3). Given that the shape and the contrast between the inner and outer parts of these eye-like parts of objects resemble those of real eyes (Burra & Kerzel, 2021.3; Epley et al., 2007.1), individuals could associate the eye-like features of objects with averted gaze, orienting their attention accordingly (Kobayashi & Kohshima, 2001; Zhang et al., 2011.1). Head orientation has been shown not only to trigger attentional shifts (Hoehl et al., 2014.2; Tautvydaite & Burra, 2024) but also to amplify the effect of an averted gaze (Jenkins & Langton, 2003.1; Kluttz et al., 2009; Qian et al., 2013.1; Schwaninger et al., 2005; Todorovic, 2006). When individuals see face-like objects, they are likely to mentally imbue these face-like configurations with social attributes, particularly those related to head orientation. This tendency raises an intriguing question: Do the attentional shifts triggered by face-like objects stem primarily from their physical properties or from the social attributions that individuals unconsciously imbue to them? Our investigation explores this dichotomy, examining whether it is the eye-like parts or the face-like configuration that primarily drives these attentional shifts.

This study employed a gaze cueing task to investigate these questions. Unlike the Posner cueing task, which uses directional arrows (Posner, 1980), the gaze cueing task is a well-established paradigm for investigating the attentional underpinnings of attentional shifts triggered by averted gazes (Edwards et al., 2015.1; Friesen & Kingstone, 1998). Typically, an averted gaze face is presented as the cueing stimulus at the center of the screen. Previous studies have used various types of averted gaze faces, including real faces (Ji et al., 2020.1; Le et al., 2021.4; Lin et al., 2020.2; Liu et al., 2021.5; McCrackin & Itier, 2018.2; van Rooijen et al., 2018.3; Vivanti et al., 2017), computer-generated faces (Caruana et al., 2019; Ji et al., 2022.4), and schematic faces (Birmingham & Kingstone, 2009.1; Dodd et al., 2011.2; Friesen & Kingstone, 1998; Wei et al., 2019.1; Yokoyama et al., 2019.2; Yokoyama & Takeda, 2019.3). After a short stimulus onset asynchrony of 200 to 350 ms (Conty et al., 2006.1; McCrackin & Itier, 2018.2; Yoxon et al., 2019.4), a geometric pattern or a light spot is presented as the target stimulus on either the left or right side of the screen (Ishikawa et al., 2021.2; Ji et al., 2020.1). Participants usually need to respond to the location of the target stimuli (Friesen & Kingstone, 1998; Frischen et al., 2007; Ishikawa et al., 2021.2; Ji et al., 2020.1, 2022.4), they can also perform detection or identification tasks (Bayliss et al., 2011.3; Burra & Kerzel, 2021.3; Friesen & Kingstone, 1998). The use of a localization task is often chosen because cueing effects tend to be larger in such tasks (McKay et al., 2021.6). Even when participants are explicitly informed that the direction of the cueing stimuli does not predict the location of the subsequent target stimuli, they still respond faster in congruent trials (where the target was presented at the pointing-to locations) than in incongruent trials (where the target was not presented at the pointing-to locations; Friesen & Kingstone, 1998). The gaze cueing effect refers to the difference in response time (RT) between congruent and incongruent trials (calculated as the mean incongruent RTs minus the mean congruent RTs; Ishikawa et al., 2021.2; Talipski et al., 2020.3). The gaze cueing effect is influenced by several factors. For instance, the cueing effect disappears when the duration of the cueing stimuli exceeds 1,000 ms, a phenomenon explained by the inhibition of return (Frischen et al., 2007; Yoxon et al., 2019.4). Additionally, some studies have found a positive correlation between the extent of gaze direction and the strength of the gaze cueing effect (Yokoyama & Takeda, 2019.3; Zhang et al., 2011.1). Consequently, researchers often control for perceived gaze direction when using the gaze cueing task (Bayliss et al., 2011.3; Qian et al., 2013.1).

In a gaze cueing task, the gaze cueing effect is a crucial measure of attentional shifts triggered by averted gazes. This effect, defined as the difference in RT between congruent and incongruent trials, is attributed to attentional shifts (Atkinson et al., 2018.4; Ji et al., 2022.4; Talipski et al., 2020.3). Notably, some studies have found no difference in cueing effects elicited by averted gaze faces and directional arrows (Bayliss et al., 2011.3; Lin et al., 2020.2; Tipples, 2002). However, other research has shown that averted gaze faces generate a stronger cueing effect than arrows, with individuals associating this asymmetry with directional information (Gregory & Jackson, 2017.1; Ristic et al., 2012). This suggests the involvement of a social-special mechanism based on social characteristics rather than a domain-general attention mechanism (Bayliss et al., 2011.3; Dalmaso et al., 2020.4). The enhanced cueing effect may stem from the heightened saliency, ecologically valid signals, or social-biological information in averted gaze faces compared to directional arrows (Birmingham & Kingstone, 2009.1; Ishikawa et al., 2021.2; Tipper & Bayliss, 2018.5). Besides, dividing attentional shifts into attentional benefit and cost can be challenging when cue-probe congruency is limited to two conditions (congruent vs. incongruent; Bayliss et al., 2011.3; Yokoyama & Takeda, 2019.3). Friesen and Kingstone (1998) used schematic faces as cueing stimuli and introduced directed gaze faces as neutral trials. They found that cue-probe congruency produced an attentional benefit without a cost and suggested that the schematic faces used as the cueing stimuli might weaken the gaze cueing effect. Subsequent research often employs real faces (Ji et al., 2020.1; Le et al., 2021.4; Lin et al., 2020.2; Liu et al., 2021.5; McCrackin & Itier, 2018.2; van Rooijen et al., 2018.3; Vivanti et al., 2017) or computer-generated faces (Caruana et al., 2019; Ji et al., 2022.4) as cueing stimuli. The gaze cueing effect can be more comprehensively elucidated by using these faces and comparing RTs across all three conditions of cue-probe congruency.

Our study explores the similarities and differences in attentional shifts induced by face-like objects and averted gaze faces. Investigating attentional shifts induced by face-like objects can reveal the underlying processing mechanisms of these objects, specifically what information individuals extract to make these shifts (Atkinson et al., 2018.4; Palmer & Clifford, 2020; Rahman & Boxtel, 2022.5). Recent studies have explored the role of face-like objects in attentional diversion. For instance, Takahashi and Watanabe (2013) used face-like objects and schematic faces as cueing stimuli and found no differences in cueing effects between them. Two potential explanations might account for their findings: First, the schematic faces lacked ecological validity compared to real faces (Friesen & Kingstone, 1998; Le et al., 2021.4; Liu et al., 2021.5; van Rooijen et al., 2018.3) or computer-generated faces (Caruana et al., 2019), potentially resulting in weaker cueing effects. Second, the study did not control gaze direction, which influenced the magnitudes of the gaze cueing effects for both face-like objects and schematic faces. Our empirical investigations for verifying these explanations proceed as follows. In a preliminary experiment, we measured the directions of averted gaze faces and face-like objects, categorizing them as two different cueing stimuli for subsequent experiments (see Supplemental Information). In Experiment 1, we used the gaze cueing task to test whether cueing effects in face-like objects would be similar to those observed in averted gaze faces. Nonetheless, the role of low-level physical properties, eye-like features, or face-like configurations in triggering attentional shifts remained undetermined. Thus, in Experiment 2, we isolated the averted gazes and eye-like parts of the two stimuli, and in Experiment 3, we presented the two stimuli in an inverted form. Finally, we compared cueing effects among three presenting modes in Experiment 4 to investigate these issues.

Chen, Z., Wen, M., Liu, X., & Fu, D. (2025).
How face-like objects and averted gaze faces orient our attention: The role of global configuration and local features. I-Perception, 16(4). https://doi.org/10.1177/20416695251352129.

Copyright: © 2025 The authors.
Published by SAGE Publications Ltd. Open access.
Reprinted under a Creative Commons Attribution 4.0 International license (CC BY 4.0)

These findings pose a significant problem for creationist arguments that rely on perceived intentionality or divine messaging in random natural patterns. When believers claim to see the face of Jesus in toast or the Virgin Mary in a tree trunk, they often cite these experiences as evidence of supernatural communication or spiritual confirmation. Yet this research demonstrates that such perceptions are the predictable outcome of how the human brain processes visual information—not a signal from beyond, but a misfiring of an evolved pattern-recognition system designed to prioritise face detection.

The study shows that our brains respond even more strongly to inanimate “faces” than to real human gaze cues, suggesting that pareidolia isn't a spiritual experience but a deeply ingrained neurological reflex. Evolutionary theory readily explains this: a brain that errs on the side of caution by seeing faces where none exist would have helped early humans survive in uncertain environments. The occasional false alarm is a small price to pay for avoiding the potentially fatal mistake of missing a real threat—or a social ally. Creationism, on the other hand, offers no coherent explanation for why a supposedly perfect designer would endow us with such a faulty perceptual system prone to false positives.

By exposing pareidolia as a by-product of evolved cognitive biases rather than divine revelation, the research undermines one of the more emotionally compelling, but scientifically empty, arguments often cited by religious believers. It moves face illusions out of the realm of the miraculous and into the domain of neuroscience and evolutionary psychology—where they belong. As with so many human experiences once thought to be supernatural, science shows that what feels profound is, in fact, perfectly natural.

Advertisement

Amazon