Pareidolia is the phenomenon of recognizing a meaningful pattern, shape, or object from ambiguous forms in visual scenes. We experience pareidolia when an object is perceived even though it is not actually there (Kato and Mugitani 2015; Liu et al. 2014; Mamiya et al. 2016; Proverbio and Galli 2016). When we recognize an animal’s form in the clouds, a pattern of a leaf in a cup of coffee, or the figure of a reaching arm in the shadow of the tree, we are experiencing pareidolia. More notable examples include seeing a man’s face on the moon or Jesus’ portrayal on a piece of toast (Liu et al. 2014). Most studies focus on a specific type of pareidolia – face pareidolia – in which faces or face-like patterns are recognized in the absence of an actual face (Liu et al. 2014; Proverbio and Galli 2016; Ryan et al. 2016; Takahashi and Watanabe 2013; Wardle et al. 2017). Although examples of face pareidolia emerge spontaneously in the environment, there are also examples of face pareidolia present in various forms of art. For example, Guiseppe Arcimboldo’s work depicts human portraits composed of various items such as fruits, vegetables, and flowers.
The experience of face pareidolia is likely caused by our sensitivity to certain categories of importance. Faces are inherently one of the most socially relevant stimuli we encounter (Palermo and Rhodes 2007). They convey socially important information such as an individual’s identity, sex, emotional state, etc. On a day-to-day basis we encounter hundreds of faces, resulting in our extreme sensitivity to facial structure, ultimately promoting the rapid detection and identification of different individuals (Crouzet 2010). Their relevance is likely the cause of our ability to so easily recognize face-like patterns even when they are not present (Ryan et al. 2016). From an evolutionary perspective, our threshold for detecting faces could be low in order to avoid any potentially dangerous situations, resulting in an increased number of false positives for seeing faces when they are not actually present (Takahashi and Watanabe 2013). Our susceptibility to such instances of face pareidolia could also be a result of our tendency to anthropomorphize objects (Proverbio and Galli 2016).
Several studies have investigated pareidolia using neurological measures such as event-related potentials (ERPs), functional magnetic resonance imaging (fMRI), or magnetoencephalography (MEG) to see whether activation resulting from illusory faces is associated with activation caused by seeing real faces. For example, the presentation of illusory faces results in activation of the fusiform face area (FFA) – an area of the brain related to face perception (Wardle et al. 2017; Liu et al. 2014). Similarly, Proverbio and Galli (2016) found that images of face pareidolia elicited ERP responses in brain areas involved in face processing, an effect that was stronger in females than in males. Finally, a study using MEG to measure the response to face pareidolia found activation in the ventral fusiform cortex, an area also activated by the perception of real faces. An important aspect to that experiment was that the presentation of illusory faces resulted in an activation pattern that was not only similar in the brain regions activated but also similar in response times from stimulus onset to activation signal. These results suggest that our perception of face pareidolia is not a post-recognition phenomenon but is rooted in the early stages of visual analysis (Hadjikhani et al. 2009).
Other research has used behavioral measures such as looking time to assess perception of illusory faces. This method builds off research indicating that humans have an innate preference for looking at faces or face-like objects (Farroni et al. 2005). In a study presenting illusory faces to infants ranging in age from 8 to 12 months, researchers found that fixations were most concentrated on the “mouth” areas of images during the simultaneous presentation of sounds (Kato and Mugitani 2015). This behavior is consistent with infants’ sound-mouth association or their tendency to look at one’s mouth as a source of sound. Infants who were 7- to 9-months old showed a preference for upright Arcimboldo images, but not their inverted counterparts, indicating that they recognize these images as faces (Kobayashi et al. 2012). Beran et al. (2017) presented three and a half to five-year-old children with Arcimboldo images after training them to categorize images of foods and faces. Their results showed that preschool-aged children were successful in categorizing this type of illusory face as a face more often than scattered images, suggesting that, even at this young age, children are susceptible to face pareidolia. And, Takahashi and Watanabe (2013) found that examples of face pareidolia elicited shifts in gaze only when the illusory face pattern was explicitly recognized. Therefore, when illusory faces were perceived, an attentional response similar to that of true faces was triggered.
Several studies, including the fMRI study discussed above (Proverbio and Galli 2016), have investigated whether there are gender differences in perceiving face pareidolia. Gender differences have previously been shown in the perception of other social signals such as reading body language (Krüger et al. 2013) or emotional valence (Kret and De Gelder 2012). For example, two studies investigating gender differences in face pareidolia found that females recognize Arcimboldo-type illusory faces more readily than males, and that females perceive these illusory faces as more likeable than male perceivers (Pavlova et al. 2016b).
Perception of face pareidolia has also been used a platform for studying specific aspects of visual processing such as form perception or the underlying mechanism of face processing. For example, humans tend to process stimuli, especially faces, in a global or holistic manner. That is, we view and discriminate objects from each other based on their overall form rather than by their individual parts. For example, Navon (1977) presented stimuli that consisted of smaller letters forming the shape of a larger letter. Adults responded faster to the global form (the larger letter) rather than the local properties (the smaller, individual letters).
Our inherent tendency to process stimuli, especially faces, holistically also makes face pareidolia an ideal platform for studying social disorders in which these perceptual processes are disrupted. For example, individuals with autism spectrum disorder (ASD), in addition to having deficits in discriminating and interpreting social stimuli, typically show a local bias in which they discriminate objects by their individual features instead of global form (Happé and Frith 2006). In general, individuals with ASD are less likely to identify pareidolic faces than typically developing controls (Pavlova et al. 2017; Ryan et al. 2016). Similarly, adults with prosopagnosia (also referred to as “face blindness”) are unable to discriminate faces from each other and have marked deficits in the holistic processing of faces (Avidan et al. 2011; Busigny et al. 2010). However, evidence suggests that despite this impairment in face processing and a bias for local processing, global processing remains intact and allows for the perception of global Navon figures as well as illusory faces (Avidan et al. 2011; Busigny et al. 2010). This is also true for patients with simultanagnosia, who have an extreme local processing bias but are still able to recognize a face-like configuration in Navon and Arcimboldo images (Dalrymple et al. 2007). Individuals with Williams syndrome – who typically exhibit greater attention to faces and social stimuli – were unable to identify faces in pareidolic images (Pavlova et al. 2016a). Thus, studying whether or not individuals with social disorders such as these perceive illusory faces (in which the local features are not face-like but the global form is face-like) allows for a greater understanding of what perceptual mechanisms underlying face processing are impaired in these populations.
Forms of pareidolia in nature may be recognizable to animals. For example, eye-spot patterns are sometimes used by prey species as a method of deterring potential predators. Although eye-spot mimicry has been shown to successfully deter predators, it is unclear whether this is due to the explicit recognition of the eye-like pattern by other animals or more simply due to the conspicuousness of the markings (Kelley and Kelley 2014).
However, comparative research has shown that many species of nonhuman animals experience various visual illusions, some in a similar manner to humans (Fujita et al. 2012; Kelley and Kelley 2014). This research has continued to explore the similarities and differences among human and nonhuman animal perception. Therefore, it is a natural progression of this field to investigate animals’ experience of the pareidolia illusion, especially of face pareidolia given a shared emphasis on the importance of face stimuli across species.
Pareidolia has only recently been studied in nonhuman animals. Faces are an important and socially relevant stimulus to many animal species. For example, multiple species show an inherent tendency to look at faces or the head regions of other individuals (Salva et al. 2011; Sugita 2008). Further, despite typically being local processors for other forms of visual information, the importance of faces is demonstrated by evidence that monkeys process faces holistically (see Parr 2011, for a review). Thus, research has investigated whether the salience of faces is strong enough to elicit this illusion as well as the holistic processing of these illusory faces.
A few recent studies have investigated whether nonhuman primates experience this illusion. In one study with rhesus monkeys and capuchin monkeys, subjects were first trained to categorize images on a computer as either faces or foods (Beran et al. 2017). After learning this task, monkeys were then presented with Arcimboldo images as well as scattered Arcimboldo portraits, real faces, and object images. Monkeys were just as likely to categorize Arcimboldo images as faces as they were to classify their scattered image counterparts as being faces. These results seem to indicate that the monkeys were unable to perceive these illusory faces, perhaps due to their tendency to process environmental stimuli in a more local manner (i.e., attending to the individual components rather than its whole). If this is the case, then monkeys would fail to recognize the face-like configuration, suggesting instead that face pareidolia is a phenomenon unique to humans.
A second study of nonhuman primates’ ability to experience face pareidolia used eye tracking to measure the gaze behavior of rhesus macaque monkeys during the presentation of illusory faces (Taubert et al. 2017). Images of conspecific faces, illusory faces, and content-matched non-face-like objects were presented in a visual paired comparison task in which monkeys were able to freely view images. Because of the monkeys’ preference for viewing faces, researchers investigated whether the looking patterns typical of viewing a true face would also be evident when the monkeys viewed images containing illusory faces. The results showed that rhesus monkeys spent more time looking at illusory faces than at objects that did not contain any face-like configuration. In addition to this finding, viewing patterns revealed an increased number of fixations on the “eye” and “mouth” regions of pareidolia images. These viewing patterns are consistent with how nonhuman primates look at images of conspecific faces (Dal Monte et al. 2015), and they suggest that nonhuman primates do possess the ability to perceive illusory faces. Finally, in a similar experiment using rhesus macaque monkeys with bilateral amygdala lesions (an area of the brain involved in face and emotional valence processing), these viewing preferences were no longer apparent, emphasizing the importance of specific brain areas in responding to faces and face-like images (Taubert et al. 2018).
More work in this area will continue to focus on the nature of nonhuman animals’ experience of this phenomenon. Specifically, the question remains as to whether the experience of this phenomenon in monkeys is more implicit (like the innate preference to look at face-like patterns) or whether nonhuman primates explicitly recognize the face-like aspects of these images.
Overall, face pareidolia is a well-established visual phenomenon in humans. The similarity between neural and behavioral responses to real and illusory faces have elucidated the value in studying inanimate objects with illusory facial structure to better understand the neural basis of face perception and its development. Studying this phenomenon in animals will inform our understanding of the mechanisms underscoring face perception in humans as well as illustrate how these mechanisms may have evolved throughout time. Current comparative research has led researchers to draw preliminary, contrasting opinions regarding the perception of illusory faces in nonhuman primates. Therefore, more work is needed to better understand the experience of animals when presented with this illusion. By learning more about illusory perception in other species, we will gain insight into the mechanisms underlying our own perception of real and illusory faces.
- Avidan, G., Tanzer, M., & Behrmann, M. (2011). Impaired holistic processing in congenital prosopagnosia. Neuropsychologia, 49(9), 2541–2552. https://doi.org/10.1016/j.neuropsychologia.2011.05.002.CrossRefPubMedPubMedCentralGoogle Scholar
- Busigny, T., Joubert, S., Felician, O., Ceccaldi, M., & Rossion, B. (2010). Holistic perception of the individual face is specific and necessary: Evidence from an extensive case study of acquired prosopagnosia. Neuropsychologia, 48(14), 4057–4092. https://doi.org/10.1016/j.neuropsychologia.2010.09.017.CrossRefPubMedGoogle Scholar
- Dalrymple, K. A., Kingstone, A., & Barton, J. J. S. (2007). Seeing trees OR seeing forests in simultanagnosia: Attentional capture can be local or global. Neuropsychologia, 45(4), 871–875. https://doi.org/10.1016/J.NEUROPSYCHOLOGIA.2006.07.013.CrossRefPubMedGoogle Scholar
- Fujita, K., Nakamura, N., Sakai, A., Watanabe, S., & Ushitani, T. (2012). Amodal completion and illusory perception in birds and Primates. In How animals see the world: Comparative behavior, biology, and evolution of vision (pp. 100–116). Oxford University Press. https://doi.org/10.1093/acprof:oso/9780195334654.003.0008.CrossRefGoogle Scholar
- Kobayashi, M., Otsuka, Y., Nakato, E., Kanazawa, S., Yamaguchi, M. K., & Kakigi, R. (2012). Do infants recognize the Arcimboldo images as faces? Behavioral and near-infrared spectroscopic study. Journal of Experimental Child Psychology, 111(1), 22–36. https://doi.org/10.1016/J.JECP.2011.07.008.CrossRefPubMedGoogle Scholar
- Kret, M. E., & De Gelder, B. (2012). A review on sex differences in processing emotional signals. Neuropsychologia, 50(7), 1211–1221. https://doi.org/10.1016/j.neuropsychologia.2011.12.022.CrossRefPubMedGoogle Scholar
- Palermo, R., & Rhodes, G. (2007). Are you always on my mind? A review of how face perception and attention interact. Neuropsychologia, 45(1), 75–92. https://doi.org/10.1016/j.neuropsychologia.2006.04.025.CrossRefPubMedGoogle Scholar
- Taubert, J., Flessert, M., Wardle, S. G., Basile, B. M., Murphy, A. P., Murray, E. A., & Ungerleider, L. G. (2018). Amygdala lesions eliminate viewing preferences for faces in rhesus monkeys. Proceedings of the National Academy of Sciences of the United States of America, 115(31), 8043–8048. https://doi.org/10.1073/pnas.1807245115.CrossRefPubMedPubMedCentralGoogle Scholar