Pareidolia: Meaning-Making in a Meaningless World Through The Lens of Signal Detection Theory (SDT)
Theo Tobel
Introduction
Imagine you’re hiking through a forest, and out of the corner of your eye, you spot a snake on the ground—only to realize that it’s a tree root. Or you’re lying on the grass on a warm summer’s day and you start seeing babies and elephants in the clouds. This psychological phenomenon is known as pareidolia, and it speaks to something about the brain’s tendency to see familiar objects within ambiguous, random data, often referred to as “noise.” One of the most prevalent and well-documented types is face pareidolia, where people see faces in toast, cereal, or even in a mountain.
Historically, pareidolia has significant cultural, artistic, and evolutionary relevance. It has been used by artists for centuries; for example, 16th-century painter Giuseppe Arcimboldo used miscellaneous objects, such as vegetables, to make portraits that were instinctively seen as human faces. It is theorized that early humans interpreted cracks in cave walls as humans and animals during the Paleolithic Age (Wisher et al., 2024). Pareidolia may have even influenced the creation of omniscient beings, such as the image of God, who is sometimes depicted as a cloud-like object (Guthrie, 1993). Pareidolia may have also given early humans a survival advantage, where detecting a face or animal in the wild in hunter-gatherer societies could be a matter of life-or-death, providing evolutionary advantages by making us more accustomed to detecting threats (Hamilton et al., 2024).
It is clear that the human brain is sensitive to faces, yet the potential mechanisms underlying this phenomenon are not well understood. Signal Detection Theory (SDT) provides a solid framework for understanding pareidolia by offering three hypotheses. Evidence for these three possible explanations will be explored through the findings in the field of cognitive neuroscience, along with suggesting how future research can further explore and elucidate the mysteries of the meaning-making brain.
Signal Detection Theory and Pareidolia
SDT is a framework used to conceptualize human decision when distinguishing between a signal and noise under uncertainty. SDT has two key components: sensitivity and response bias, which is also called the criterion. Sensitivity measures the difference in standard deviations between the signal (e.g., a face) and noise (e.g., a pancake resembling a face) distributions. High sensitivity occurs when the two distributions are clearly distinct, for example, when a stimulus is unmistakably a human face. The second component, response bias, represents the threshold value an individual uses to decide whether a stimulus is noise or signal. If mistaking a pancake for a face carries little consequence, the response bias is low, or “liberal.”
In the context of pareidolia, when people perceive faces in uncertain stimuli, this can be understood as a false alarm. A false alarm is one of four outcomes in a SDT decision matrix, and describes a scenario in which an individual thinks that there is a face in an environment and there is in reality no face present. People may exhibit higher levels of pareidolia for three reasons according to SDT: (1) they possess lower sensitivity, meaning that they have trouble distinguishing between faces and non-faces; (2) they have a less “conservative” response bias, meaning that they trade hits for false alarms; (3) both factors could be at play through bottom-up and top-down processing.
The Cognitive Neuroscience of Pareidolia
The field of cognitive neuroscience can provide insights into the mechanisms that support the different SDT-based hypotheses. The first hypothesis—that individuals who experience pareidolia have reduced sensitivity—is supported by studies that have found that individuals with schizophrenia (SZ) have a lower sensitivity to faces. SZ is a mental disorder characterized by significant impairments in the perception of reality (NIMH, n.d.). In these individuals, there may be more overlap between the two distributions of noise and face stimuli, though a consensus has not been reached in the literature (Romagnano et al., 2022; Rolf et al., 2020; Mavrogiorgou et al., 2021). The ventral occipital cortex (VOT) which is involved in bottom-up processing—a type of processing that relies on sensory input rather than prior expectations—influences sensitivity in detecting local facial features (Liu et al., 2014). In individuals with SZ, this region may be disrupted, impairing the ability to accurately distinguish faces from noise.
The second hypothesis that a more liberal response bias leads to increased pareidolias can be linked to the top-down activation of the right fusiform facial area (rFFA), which is specifically tuned to face and illusory face perception, by the prefrontal cortex (PFC). Liu et al. (2014) discovered that during pareidolia, higher-order regions like the PFC modulate the rFFA, increasing its likelihood of being activated by noise. In SDT terms, this corresponds to a lower response bias because it lowers the decision threshold.
Lastly, both reduced sensitivity and a liberal response bias could influence pareidolia through the combination of bottom-up and top-down processing. For example, Zhou & Meng (2020) suggested that sex differences in face perception could be explained by lower sensitivity to facial features and a top-down bias toward perceiving faces. Hadjikhani (2009) showed that event-related potential responses to real and illusory faces were similar in the N170 component. Studies such as these further emphasize the role of both perceptual mechanisms and cognitive expectations in shaping false face detections.
In all, there is evidence from research on the neural correlates of pareidolia that strongly supports each of the three SDT-based hypotheses, but more studies are needed to understand whether pareidolia can be explained fully by the SDT framework.
Future Research Directions
Future research can employ techniques from neuroeconomics, such as reward-punishment paradigms, to manipulate response bias in face detection tasks by incentivizing particular responses. This approach would test the hypothesis that response bias in face detection contributes to the occurrence of pareidolia. Researchers could examine how penalizing pareidolic responses influences the neural mechanisms of face detection. In the vein of typical neuroeconomics measurements, future papers could investigate eye-tracking and response time to advance understanding for how humans distinguish the two distributions of “not face” and “face” in the context of pareidolia. Another route could be using noninvasive brain stimulation to temporarily lesion the neural circuits involved in pareidolia. For example, cognitive neuroscientists could intentionally disrupt the PFC-FFA pathway in order to elicit causal evidence for the role of higher systems in pareidolia.
Conclusion
Pareidolia provides a fascinating example of how our brains create meaning from random stimuli, and face pareidolia highlights the importance of face recognition for humans. SDT offers a model for understanding this phenomenon, and although it is simplified, it helps illustrate the brain’s use of top-down and bottom-up processes. Pareidolia is especially exciting because it is not fully understood, presenting a ripe area for future research with neuroeconomics-inspired experiments and brain stimulation to uncover causal mechanisms.
About the Author
Theo Tobel ('27) is a sophomore at Harvard College, studying Philosophy and Neuroscience.
References
Imagine you’re hiking through a forest, and out of the corner of your eye, you spot a snake on the ground—only to realize that it’s a tree root. Or you’re lying on the grass on a warm summer’s day and you start seeing babies and elephants in the clouds. This psychological phenomenon is known as pareidolia, and it speaks to something about the brain’s tendency to see familiar objects within ambiguous, random data, often referred to as “noise.” One of the most prevalent and well-documented types is face pareidolia, where people see faces in toast, cereal, or even in a mountain.
Historically, pareidolia has significant cultural, artistic, and evolutionary relevance. It has been used by artists for centuries; for example, 16th-century painter Giuseppe Arcimboldo used miscellaneous objects, such as vegetables, to make portraits that were instinctively seen as human faces. It is theorized that early humans interpreted cracks in cave walls as humans and animals during the Paleolithic Age (Wisher et al., 2024). Pareidolia may have even influenced the creation of omniscient beings, such as the image of God, who is sometimes depicted as a cloud-like object (Guthrie, 1993). Pareidolia may have also given early humans a survival advantage, where detecting a face or animal in the wild in hunter-gatherer societies could be a matter of life-or-death, providing evolutionary advantages by making us more accustomed to detecting threats (Hamilton et al., 2024).
It is clear that the human brain is sensitive to faces, yet the potential mechanisms underlying this phenomenon are not well understood. Signal Detection Theory (SDT) provides a solid framework for understanding pareidolia by offering three hypotheses. Evidence for these three possible explanations will be explored through the findings in the field of cognitive neuroscience, along with suggesting how future research can further explore and elucidate the mysteries of the meaning-making brain.
Signal Detection Theory and Pareidolia
SDT is a framework used to conceptualize human decision when distinguishing between a signal and noise under uncertainty. SDT has two key components: sensitivity and response bias, which is also called the criterion. Sensitivity measures the difference in standard deviations between the signal (e.g., a face) and noise (e.g., a pancake resembling a face) distributions. High sensitivity occurs when the two distributions are clearly distinct, for example, when a stimulus is unmistakably a human face. The second component, response bias, represents the threshold value an individual uses to decide whether a stimulus is noise or signal. If mistaking a pancake for a face carries little consequence, the response bias is low, or “liberal.”
In the context of pareidolia, when people perceive faces in uncertain stimuli, this can be understood as a false alarm. A false alarm is one of four outcomes in a SDT decision matrix, and describes a scenario in which an individual thinks that there is a face in an environment and there is in reality no face present. People may exhibit higher levels of pareidolia for three reasons according to SDT: (1) they possess lower sensitivity, meaning that they have trouble distinguishing between faces and non-faces; (2) they have a less “conservative” response bias, meaning that they trade hits for false alarms; (3) both factors could be at play through bottom-up and top-down processing.
The Cognitive Neuroscience of Pareidolia
The field of cognitive neuroscience can provide insights into the mechanisms that support the different SDT-based hypotheses. The first hypothesis—that individuals who experience pareidolia have reduced sensitivity—is supported by studies that have found that individuals with schizophrenia (SZ) have a lower sensitivity to faces. SZ is a mental disorder characterized by significant impairments in the perception of reality (NIMH, n.d.). In these individuals, there may be more overlap between the two distributions of noise and face stimuli, though a consensus has not been reached in the literature (Romagnano et al., 2022; Rolf et al., 2020; Mavrogiorgou et al., 2021). The ventral occipital cortex (VOT) which is involved in bottom-up processing—a type of processing that relies on sensory input rather than prior expectations—influences sensitivity in detecting local facial features (Liu et al., 2014). In individuals with SZ, this region may be disrupted, impairing the ability to accurately distinguish faces from noise.
The second hypothesis that a more liberal response bias leads to increased pareidolias can be linked to the top-down activation of the right fusiform facial area (rFFA), which is specifically tuned to face and illusory face perception, by the prefrontal cortex (PFC). Liu et al. (2014) discovered that during pareidolia, higher-order regions like the PFC modulate the rFFA, increasing its likelihood of being activated by noise. In SDT terms, this corresponds to a lower response bias because it lowers the decision threshold.
Lastly, both reduced sensitivity and a liberal response bias could influence pareidolia through the combination of bottom-up and top-down processing. For example, Zhou & Meng (2020) suggested that sex differences in face perception could be explained by lower sensitivity to facial features and a top-down bias toward perceiving faces. Hadjikhani (2009) showed that event-related potential responses to real and illusory faces were similar in the N170 component. Studies such as these further emphasize the role of both perceptual mechanisms and cognitive expectations in shaping false face detections.
In all, there is evidence from research on the neural correlates of pareidolia that strongly supports each of the three SDT-based hypotheses, but more studies are needed to understand whether pareidolia can be explained fully by the SDT framework.
Future Research Directions
Future research can employ techniques from neuroeconomics, such as reward-punishment paradigms, to manipulate response bias in face detection tasks by incentivizing particular responses. This approach would test the hypothesis that response bias in face detection contributes to the occurrence of pareidolia. Researchers could examine how penalizing pareidolic responses influences the neural mechanisms of face detection. In the vein of typical neuroeconomics measurements, future papers could investigate eye-tracking and response time to advance understanding for how humans distinguish the two distributions of “not face” and “face” in the context of pareidolia. Another route could be using noninvasive brain stimulation to temporarily lesion the neural circuits involved in pareidolia. For example, cognitive neuroscientists could intentionally disrupt the PFC-FFA pathway in order to elicit causal evidence for the role of higher systems in pareidolia.
Conclusion
Pareidolia provides a fascinating example of how our brains create meaning from random stimuli, and face pareidolia highlights the importance of face recognition for humans. SDT offers a model for understanding this phenomenon, and although it is simplified, it helps illustrate the brain’s use of top-down and bottom-up processes. Pareidolia is especially exciting because it is not fully understood, presenting a ripe area for future research with neuroeconomics-inspired experiments and brain stimulation to uncover causal mechanisms.
About the Author
Theo Tobel ('27) is a sophomore at Harvard College, studying Philosophy and Neuroscience.
References
- Guthrie, S. E. (1993). Faces in the clouds: A new theory of religion. Oxford University Press.
- Hadjikhani, N., Kveraga, K., Naik, P., & Ahlfors, S. P. (2009). Early (M170) activation of face-specific cortex by face-like objects. Neuroreport, 20(4), 403–407. https://doi.org/10.1097/WNR.0b013e328325a8e1
- Hamilton M, Stent S, DuTell V, Harrington A, Corbett J, Rosenholtz R, Freeman WT. Seeing faces in things: A model and dataset for pareidolia. arXiv:2409.16143 [Preprint]. September 24, 2024. Available from: https://doi.org/10.48550/arXiv.2409.16143.
- Liu, J., Li, J., Feng, L., Li, L., Tian, J., & Lee, K. (2014). Seeing Jesus in toast: neural and behavioral correlates of face pareidolia. Cortex; a journal devoted to the study of the nervous system and behavior, 53, 60–77. https://doi.org/10.1016/j.cortex.2014.01.013
- Mavrogiorgou, P., Peitzmeier, N., Enzi, B., Flasbeck, V., & Juckel, G. (2021). Pareidolias and Creativity in Patients with Mental Disorders. Psychopathology, 54(2), 59–69. https://doi.org/10.1159/000512129
- Rolf, R., Sokolov, A. N., Rattay, T. W., Fallgatter, A. J., & Pavlova, M. A. (2020). Face pareidolia in schizophrenia. Schizophrenia research, 218, 138–145. https://doi.org/10.1016/j.schres.2020.01.019
- Romagnano, V., Sokolov, A.N., Steinwand, P. et al. Face pareidolia in male schizophrenia. Schizophr 8, 112 (2022). https://doi.org/10.1038/s41537-022-00315-y
- U.S. Department of Health and Human Services. (n.d.). Schizophrenia. National Institute of Mental Health (NIMH). https://www.nimh.nih.gov/health/statistics/schizophrenia
- Wisher, I., Pettitt, P., & Kentridge, R. (2024). Conversations with Caves: The Role of Pareidolia in the Upper Palaeolithic Figurative Art of Las Monedas and La Pasiega (Cantabria, Spain). Cambridge Archaeological Journal, 34(2), 315–338. doi:10.1017/S095977432300028