Mystery and perception
Pareidolia and a sensed presence: what distinguishes them
A silhouette may look like someone before we know what produces it. Four investigations help to separate form, expectation and explanation, without turning a possibility into a verdict.

Imagine entering a dim room and seeing a figure by a door: apparent head and shoulders. Switching on the light reveals a jacket on a chair. This invented scene asks precisely what changed. The object may have stayed still; available information and identification changed.
If you said that "there was someone", you would mix two things: the impression of a human form and the existence of a person. If you said that "nothing happened", you would erase the initial impression. Between the two formulations there is a more useful description: "I saw an outline that I interpreted as a person; by illuminating it I identified a jacket".
A face can appear before the object
Face pareidolia means seeing a face in something else: holes, fruit or a surface. Susan Wardle and colleagues compared human faces, face-like objects and similar objects without that appearance. Their 2020 study combined functional MRI and magnetoencephalography. An initial face-like response shifted towards object representation within about 250 milliseconds. This supports rapid detection without requiring slow conscious reinterpretation. Study of Wardle and collaborators.
This offers an important distinction: to recognize the object afterwards does not prevent us from continuing to appreciate its resemblance with a face. We can know that we look at a plug and, at the same time, see an expression. The resemblance is not a proof of intent, emotion or identity. Nor does that experiment on faces show what happened in every human shadow account: its stimuli and its question were more specific.
What we expected to find
Let's add a variant to the imaginary room. Before entering, someone tells you that he has left the coat by the door. In another version he warns you that he has seen a person there. The words propose different candidates for the same silhouette. It is not enough to imagine this difference to affirm how much it would influence you; it does allow designing a comparison that distinguishes the image from the previous information.
Philipp Sterzer, Chris Frith and Predrag Petrovic investigated in 2008 how experimentally induced expectations modified the perception of an ambiguous visual movement. They found an effect of those expectations on the perceived. The result demonstrates an influence on the studied conditions, not that we can see at will anything or that all discrepancies between observers come from suggestion. Summary of the study at the University of Basel.
With an unsettling photograph, examine the image before its caption. This is an observation exercise, not a clinical experiment: note shapes, light, dark and what lies outside the frame. Then read the account. If a figure now appears, record when that interpretation arose. Recording context does not remove its influence; it prevents hiding it.
We also leave things out
Observing is not just about adding meaning. It also involves selecting. In the study published by Daniel Simons and Christopher Chabris in 1999, the participants were following basketball passes while an unexpected figure was crossing the scene. One of the variants included a person disguised as a gorilla. Part of the observers did not notice that event. Detection varied with the conditions of the task and the characteristics of what was attended. Full article by Simons and Chabris.
This differs from seeing a face in an object. There, a configuration acquires facial identity; here, something present goes unnoticed during another task. Combining them as 'the brain deceives us' loses their different questions. Ask what we interpret and what we seek: face, exit, movement or a date at the edge.
Feeling a presence is another question
In 2014, Olaf Blanke and his collaborators studied experiences of presence in neurological patients and developed a robotic test with healthy participants. A conflict between movements and tactile signals of the body itself allowed to induce sensations of presence in controlled conditions. Summary of the original study.
The finding identifies an experimental way to produce such an experience. It does not allow to diagnose the person who tells something similar, nor to retrospectively attribute all the accounts to that mechanism. A conversation, a photograph and a laboratory test offer different evidence. Treating them as interchangeable converts a possible explanation into a certainty that the available data do not support.
Looking again
Let's go back to the jacket. A useful review needs a question that can change the answer. "Is it disturbing?" can remain so after the light is lit. "What object produces this edge?" admits checking the position of the chair, sleeve and lamp. They are different questions and can coexist.
For an ambiguous scene, write three sentences. Describe: 'an elongated dark area beside the frame'. Propose: 'it could be clothing or a person'. Request evidence: 'I need different lighting or a wider view'. Keep 'head', 'gaze' and 'lurking' out of description while they remain interpretations.
Then change one condition at a time where possible: lighting, position or crop. Preserve evidence against your first reading too. A second photograph may record another instant rather than prove the first scene's circumstances. If only one small image survives, responsible identification may be impossible.
That uncertainty does not empty the experience. It makes it a better delimited question. Shadow remains interesting when we can explain how we examine it, what changed when we looked again and what detail is still missing.
Sources and references
- Wardle, S. G., Taubert, J., Teichmann, L. and Baker, C. I. (2020). Rapid and dynamic processing of face pareidolia in the human brain. Nature Communications, 11, 4518.
- Sterzer, P., Frith, C. and Petrovic, P. (2008). Believing is seeing: expectations alter visual awareness. Current Biology, 18(16), R697–R698.
- Simons, D.J. and Chabris, C.F. (1999). Gorillas in our midst: sustained inattentional blindness for dynamic events. Perception, 28, 1059–1074.
- Blanke, O. and collaborators (2014). Neurological and robot-controlled induction of an apparition. Current Biology, 24(22), 2681–2686.