Why Do We See Faces in Objects?

Two Windows and a Door… Then a Face

In an imagined scene, a woman walks through an old alley and stops before a small house. Two dark windows sit above an arched doorway. Raising her phone, she laughs: “It looks surprised to see me.” She does not think the house alive or expect an answer. She knows its windows, door, and paint, yet sees a face she cannot ignore. Sending the picture to a friend, “Look at its face” is enough to turn conversation about a building into conversation about an astonished creature.

Two different realities occupy the image: a measurable architectural arrangement and a facial configuration appearing to the viewer. No artist’s trick or strange belief is needed. Something similar happens when a suitcase looks sad, a car’s front frowns, or two knots in wood become eyes. The surprise is not that we can invent stories about objects, but that a story sometimes begins as a visual impression before we decide to tell it.

Perception research calls this “face pareidolia”: perceiving a face in something that is not a face. We limit discussion to this everyday experience involving actual objects and pictures. We diagnose nobody from it or equate seeing resemblance with believing someone exists inside an object. It opens a more precise question: how does vision find a familiar form in different material, and why does knowledge of that material not easily erase resemblance?

This small doorway opens onto perception’s relationship with knowledge, and our responsibility when turning an impression into a report. Enjoying a face in clouds differs from treating resemblance as evidence for a story about the world. Between them is room for science, art, and reflection, without one needing to displace another.

Questions to Begin the Reading

• What makes an arrangement of windows or holes look like a face?
• Do face-perception mechanisms respond to an illusory face in an inanimate object?
• Why does neural similarity not mean real and imagined faces are identical?
• How can we enjoy resemblance while preserving accurate judgement?

1. What Do We See When We Say “That Is a Face”?

Return to the house. Asked to identify the eyes, the photographer points to the windows. Asked what they actually are, she agrees with an engineer: windows. The descriptions differ not over stone placement but in level. One concerns building parts and functions, the other how those parts organise into a form in our vision. The latter does not cancel the former or alone add a human being to the scene.

This distinction prevents a common linguistic confusion. “Illusion” may suggest the entire picture is fabricated, though windows and shadows exist. Taken literally, the error lies in classifying their arrangement as a real face. “I see it as resembling a face,” however, may accurately describe experience. We need not deny experience to correct an inference loaded onto it.

Explaining his research, Colin Palmer notes that faces vary, but the arrangement of eyes and mouth offers a familiar basic configuration; a similar arrangement elsewhere may recruit face-related processing. This simplifies the phenomenon without claiming the mind holds one rigid drawing against which everything is matched. Palmer also recalls the distinction between knowledge of an object and what we see in it:

Explore the idea descriptively with paper and three small pieces: place two above the third, then alter distances and positions. One arrangement may suggest a face, another not. This observational activity does not independently reveal a neural law; lighting, experience, and expectations are uncontrolled. It makes the question tangible: which relationship between parts supported my description? Does resemblance lie in separate pieces or their arrangement?

Detecting a face differs from recognising its owner. “I see a face there” does not answer “Who is it?” Direction or expression may be described before identity. Combining these tasks into one absolute ability, then treating success in one as proof of the rest, is mistaken. A car image gives it no personality, and suggestive features reveal no driver’s intention.

This offers a more precise response to a child seeing a face in a bag. Instead of an absolute “That is wrong,” say, “I understand where you saw eyes: those are two clasps, and the zip looks like a mouth.” Dialogue first shares description of what appeared, then distinguishes the object’s nature. Wonder is not insulted, nor is the bag granted life because it resembles something living.

2. What Does Experiment Add to Wonder?

An amusing image raises a question without settling its mechanism. Testing whether resemblance recruits some face processing needs more than asking whether people see a face. In Susan Wardle and colleagues’ 2020 study, images of face-like objects were compared with similar objects lacking that resemblance and with human faces. Two experiments used functional magnetic resonance imaging and magnetoencephalography, with analyses including 16 and 22 participants respectively. The comparison matters: comparing only a bag with a human face cannot separate resemblance from the whole difference in object type.

Resemblance-related responses appeared in face-selective visual areas. Temporally, face-like objects’ response patterns were initially relatively closer to human faces, then became closer to ordinary objects around a quarter-second later. This concerns activity patterns under experimental conditions, not a clock specifying when a face leaves every viewer’s awareness; perceived resemblance may persist. The researchers describe results as consistent with a broadly responsive detection mechanism that:

Sensitivity here means readiness to detect a potential signal; selectivity means accurately distinguishing category members from nonmembers. A hypothetical example illustrates the trade-off: asking someone to call whenever they glimpse a possible face differs from asking them to call only after complete certainty. This explains a task criterion, not a claim that one brain device receives literal verbal orders this way.

Another experimental approach was Palmer and Clifford’s 2020 perceptual-adaptation study. They repeatedly showed objects with features suggesting gaze in one direction, then measured judgements of human faces’ gaze direction. Exposure biased those judgements. Removing face-like features from the object images left a smaller effect; the finding was not simply effect versus none. This supports shared aspects of gaze-processing mechanisms beyond general directional cues, without proving all processing identical.

The studies ask complementary questions: one tracks activity patterns across space and time; the other tests exposure’s effect on a later perceptual judgement. Brain imaging does not directly read thoughts, nor does changed judgement prove a bag became a social being. Research’s strength lies in narrowing alternatives, not turning a brain image into a seal authorising every attractive explanation.

Nor should eye and brain become opponents, one faithful and the other deceptive. The familiar expression may work as a joke but confuses the scientific question. Visual experience is what we seek to understand, including successful discrimination, errors, and limits. Something appearing face-like while remaining known as an object may involve two coexisting descriptions, not a battle in which falsehood defeated truth.

3. Does Every Face Capture Our Attention?

It is easy to jump from “processed as face-like” to “always attracts attention faster than everything else.” The latter needs an attention and search experiment, not a borrowed neural-pattern result. Finding something among images depends on the target and comparisons. Ask about the comparison before celebrating “faster.”

In Robert Keys, Jessica Taubert, and Susan Wardle’s 2021 study, searching for illusory faces in objects was faster than searching for matched objects in two conditions with identical or varied distractors. Yet searching for human faces was faster and more efficient than searching for objects whether face-like or not. The researchers’ abstract states the first finding:

In 2024, Lizzie Collyer and colleagues asked what happens with a changed comparison set. Three experiments compared real and illusory faces, diverse objects, and objects from one category—flowers. Real faces retained the best search performance. Illusory faces’ advantage appeared against diverse objects but not the single-category set. The team therefore described:

This comparison denies neither faces seen in objects nor adaptation or imaging results. It limits a broader generalisation. “Some processing is shared” differs from “Attention is captured identically.” Research should not be forced to choose between an exceptionless universal explanation and knowing nothing.

A Serious Objection: Were These Images Not Selected by Researchers?

Yes, and that is part of inference’s limits. Laboratory images are not a random sample of everything in the street, nor is a specified task a full day’s life. Palmer and Clifford also state that their design and analyses were not preregistered. This methodological detail calls for weighing evidence; alone, it does not invalidate it. Presenting a study’s limits alongside its result is better than hiding them or using them to reject every experiment.

Different comparisons teach us to ask the omitted scientific-news question: faster than what, for whom, under which conditions? If a headline says the brain is “programmed to see faces everywhere,” we need neither adopt it literally nor immediately accuse deception. Return to what was actually measured. An evolutionary account may be a possible explanation of face sensitivity’s usefulness, kept separate from these experiments’ direct findings; they do not replay ancestral history before measuring instruments.

4. From Resemblance to Story: Where Does Judgement Begin?

After explaining some mechanisms, a human question remains beyond imaging: what do we do with what we saw? Consider three increasingly strong statements: “There are two dark patches on the wall,” “They look like eyes to me,” and “Someone is watching me.” The first describes examinable parts, the second reports perceived resemblance, the third claims an agent and intention. Moving between them is no innocent stylistic change; each adds something requiring different evidence.

This division is an editorial thinking tool, not a new experimental brain model. It respects a speaker’s honesty without accepting every interpretation. Someone may truthfully report seeing a face-like form without that form proving their story. We need neither accuse lying nor accept a causal relationship added to sensation.

Imagine a circulating rock photograph. Two people agree it resembles a face but disagree on meaning: one sees a photographic opportunity, another a message directed personally at them. Agreement about shape adds no evidence of a message. That requires evidence about sender, means, and meaning, not repeated description of eyes and mouth. Their disagreement need not concern visual ability; it may concern rules of inference.

In art, we may intend the story from the beginning. Giving a house dialogue in a children’s tale announces imaginative play. Metaphor should not be judged as factual reporting. Trouble begins when an image moves unmarked from play to proof: aesthetic expression becomes apparent evidence, changing a writer’s or photographer’s duty towards the audience.

Designers can use face-like configurations on a façade or cover but should test the intended communication rather than assume one universal response. Does a reader see windows or eyes? Is the intended meaning amusement or surprise? These concern a particular product and audience, not ready-made answers laboratory studies give every poster designer. This article’s accompanying cover is a generated illustration, not a building photograph offered as evidence or a stimulus used in the cited studies.

Careful language also matters before circulating images. “The angry rock” works as an artistic title; “A rock resembling an angry face” is clearer when explaining the phenomenon. Adding “I see” locates the experience without diminishing it. Precision is not expression’s enemy: it tells readers when we sketch an impression and when we ask them to believe a worldly claim.

A reader may object that this scrutiny spoils a beautiful moment. Distinguish two settings. Laughing with a friend needs no research report every time, but publishing an image with a scientific or religious explanation or an accusation changes the setting. Verification matters more as more consequences are loaded onto the scene. Amusement can remain light provided it does not support a judgement beyond its capacity.

In the Light of the Quran and Islamic Wisdom

Considering perception leads to reflection on knowledge’s blessing and responsibility for its means. Allah says:

The verse recalls humanity’s original state and given means of learning, directing towards gratitude. Abd al-Rahman al-Saadi links hearing, sight, and hearts’ particular mention to their role in acquiring knowledge, explaining gratitude as using these gifts in obedience to Allah. He describes their standing with:

Our proposed connection reflects on gratitude in knowing: using sight, learning to examine what we see, and distinguishing observation from interpretation. This draws on the article’s subject, not an exclusive interpretation of the verse or a claim that it describes face-pareidolia experiments or brain regions. Nor should its word order become a neural diagram merely because it suits a modern presentation we admire.

These limits encourage fruitful humility. Acknowledging that vision detects resemblance does not deny sight’s value. Verification need not show ingratitude; it may be good use of the gift. A blessed instrument does not make every judgement through it infallible, just as corrigible judgement does not make the instrument worthless.

Contemplating a face in a cloud, we can thank Allah for creation’s beauty and reflection’s breadth without attributing an unsupported personal message to the cloud. Preserving belief need not dry up imagination. We should know when we reflect and when we attribute a report to Allah or a creature. The latter is a weightier claim; visual resemblance alone supplies insufficient grounds.

Another courtesy concerns addressing others. If a relative shares a picture with meaning we do not see, first understand what they indicate, then explain resemblance’s difference from a claim. Mockery adds no evidence to our position; accepting the inference out of politeness may preserve confusion. We can calmly say, “I see the resemblance you mean, but that interpretation needs something more.”

Religious texts thus retain their place, experimental findings theirs, and reflection is declared as reflection. Combining them does not require fusing them into one statement, but asking what each offers: experiments test aspects of perception, commentary explains scripture in context, reflection asks about the manners of knowledge and action. Distinction enables clear dialogue.

5. Exercises That Preserve Wonder and Discipline Inference

Home need not become a laboratory to learn from this phenomenon, but name the activity correctly. These suggestions are observation, conversation, and writing exercises, not intelligence tests or diagnosis. The earlier studies do not show that doing them improves a particular ability. They make the distinctions discussed usable in everyday situations.

A Notebook With Two Levels of Description

Choose an ordinary object’s image that looks face-like to you. First describe it without eyes, mouth, or expression: “Two rectangular openings above a dark arch.” Then describe experience: “It looks surprised to me.” Place them side by side. What did the second add: a part physically present, or human organisation of existing parts? Keep both answers instead of rushing to cancel one with the other.

Add a third line for evidence needed to go beyond resemblance. If saying “The house’s owner intentionally designed the windows as eyes,” what supports that intention? You may need a statement, plan, or document. Resemblance alone does not establish purpose. The line separates design’s effect on viewers from its maker’s intention, whether the work is a building, image, or household object.

One Change Instead of a New Story

In your own drawing, move one window or cover part of the shape, then describe remaining resemblance. Do not change lighting, angle, and positions together if investigating one particular change. The face may persist or weaken; record your experience without attributing it to everyone. This practises questions and restrained inference; it measures no particular neural pathway.

Let another participant describe the image before hearing your description. The aim is not proving speech caused everything they see, but preserving observation order: what they said before a hint and added after. Recording an event differs from explaining its cause here too. Failure to see a face is no deficiency, nor rapid detection a certificate of superiority; the activity was not designed for those judgements.

A Photo Walk Without Intrusion

Families can gather examples from their own possessions: a bag, jug, or cupboard door. Describe the features suggesting resemblance, then propose an explicitly artistic title. Outside images require respecting people’s privacy and photographers’ rights, and attributing work with a source. Amusement does not automatically permit republication, nor does online visibility make rights unknown or void.

With children, allow several descriptions: “I see it sleeping,” “I see it surprised.” Ask which part supported each. Instead of competing to guess the correct answer, name the sign each child relied on. End by returning the object to its function: the bag carries books, even if its zips gave us an amusing story this morning.

Editing an Image Before Sharing

Before publishing an image and caption, check three things: did you describe resemblance or claim an event? Did you distinguish an original image from drawing or modification? Does the caption need an external source? Disclose generation or editing when omission would mislead. Link scientific explanations to the study supporting the specific claim, not a scientific page that does not say what you attribute to it.

Avoid titles turning wonder into a reader test: “Seeing this means you are a genius,” or “Only people with this trait will see it.” Without a valid test and fitting evidence, the title adds a judgement the picture cannot supply. Simpler wording can attract honestly: “Which arrangement made you see a face?” It invites observation instead of ungrounded classification.

Reading Scientific News to the End

If an image sparks research curiosity, begin with the tested question. Record the method, comparison, and one inferential limit. Do not borrow a prestigious technique’s name, such as MRI, to lend an idea authority. For example, summarise our distinction: one study compares activity patterns, another changes in gaze judgements, another time to find a target. Each opens a different window on the issue.

This care does not require every reader to reanalyse researchers’ data. Match confidence to what was read: if only the abstract, attribute what it presents without claiming review of unavailable details. If later work qualifies the finding, update the wording. Honesty is not knowing everything; it is refusing to conceal knowledge’s limits behind a stronger declaration.

Conclusion: Seeing Resemblance, Reviewing Meaning

Return to the woman before the imagined house: she can keep the picture and laughter without believing the wall an astonished being. Knowing windows are windows did not remove enjoyment; it gave a new question about how the world appears. That question begins curiosity deeper than collecting odd pictures.

The research presented indicates some overlap between real-face perception and face-like forms in objects, with limits revealed by changed tasks and comparisons. Our proposed reflective lesson is no additional laboratory result: distinguish experience’s clarity from an argument’s strength. Resemblance may be very clear while the story built upon it remains poorly evidenced.

Precision does not ask us to pass the house without imagination. It asks us to name the activity: here we observe, here imagine, here judge. Knowing the difference lets us give objects stories in art, restore their limits in knowledge, and thank the gift of sight by contemplating what it reveals and reviewing what we added.

Questions for Further Reflection

• What recently looked like a face to you, and which parts created the resemblance?
• How can you describe the experience without attributing intention or feeling to the object?
• What additional evidence do you need before making a claim from visual resemblance?
• How would you explain to a child the difference between imaginative play and factual reporting?