Figure recognition and visual attention patterns behind the observation of Palaeolithic art

RESEARCH PAPER Figure recognition and visual attention patterns behind the observation of Palaeolithic art Marı ´a Silva-Gago .Marcos Garcı ´a-Diez .Emiliano Bruner .Luis M. Martı ´nez . Felipe Criado-Boado Received: 31 March 2025 / Revised: 15 May 2025 / Accepted: 22 May 2025 ÓThe Author(s) 2025 Abstract Art visual processing is a complex cognitive process that involves perception, attention and decision-making. The initial stage of this process is characterized by visual exploration through rapid eye movements which can provide insights into attentional patterns and information processing and can be examined using eye-tracking technology. In this study, we employ eye-tracking to investigate visual perception during the observation of Palaeolithic rock art mostly from the Cantabrian seaboard, aiming to investigate the behavioural mechanisms underlying figure recognition and attentional distribution. This study comprises two experiments—one including entire panels and the other focusing exclusively on single figures—both yielding similar results. Our findings indicate that attention was predominantly directed towards the depicted figures, particularly their heads, rather than other elements within the visual field. Additionally, incomplete figures were perceived as complete, suggesting that figure recognition were influenced by Gestalt principles, internal cognitive models, and other top-down systems. Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/ s41809-025-00170-0. M. Silva-Gago (&)F. Criado-Boado Instituto de Ciencias del Patrimonio (INCIPIT), CSIC, Santiago de Compostela, Spain e-mail: [email protected] M. Garcı ´a-Diez Department of Prehistory, Ancient History and Archaeology, Universidad Complutense, Madrid (UCM), Madrid, Spain M. Garcı ´a-Diez Institut Catala `de Paleoecologı ´a Humana I Evolucio ´ Social (IPHES), Tarragona, Spain E. Bruner Museo Nacional de Ciencias Naturales (MNCN), CSIC, Madrid, Spain E. Bruner Reina Sofia Foundation Alzheimer Centre, CIEN Foundation, ISCIII, Madrid, Spain L. M. Martı ´nez Instituto de Neurociencias de Alicante (IN), CSIC-UMH, Alicante, Spain 123 J Cult Cogn Sci https://doi.org/10.1007/s41809-025-00170-0(0123456789().,-volV)(0123456789().,-volV) Keywords Eye-tracking Perception Gestalt  Rock art Vision Cognitive archaeology Graphic representation Introduction Palaeolithic rock art, as a research category for over 125 years, represents an early manifestation of human creativity and graphic ability, originating at least 65,000 years ago (Lorblanchet & Bahn, 2017; Hoffman et al., 2018). This graphic representation, derived from the replication or reinterpretation of material reality and the development of new representative means of communication. Art, as pictorial communication, constitutes a special category of images; and examining visual art can provide insights into the mechanisms of human vision and perception (Graham & Redies, 2010). Generally, in the visual arts, the creators lead the viewer’s gaze in a given order that they have planned using the composition of the work (Beelders & Bergh, 2020). We could infer this intention to Palaeolithic art, where graphic representations embody a form of human product through which the creator, whether acting individually or as a representative of the social group, transmits ideas or concepts that transcend simple visual materialisation (Pinna & Reeves, 2009). Viewing art is a complex process that involves perception, attention and decision-making, among other cognitive and affective abilities (Bailey-Ross et al., 2019; Massaro et al., 2012). The formal characteristics of the representation elicit a behavioural response in observers that has been hypothesised to be dissimilar at different periods in prehistoric art (Criado-Boado & Romero, 1993). Palaeolithic art, in particular, constitutes a structured graphic language, designed to be perceived and interpreted by members of the original human group, with the potential for a lasting impact beyond the moment of creation (Hodgson, 2003). This study focuses on this observer’s behavioural response. However, the first process of art perception involves visual scanning through fast eye movements, known as saccades, which are produced between small pauses called fixations (Kowler, 2011). The fixation-saccade system provides insights into patterns of attention and information processing and can be analysed by eye tracking technology (Carrasco, 2011). Fixation patterns reflect overt attention, which is related to directing the eyes actively towards a specific location in space, although information can also be gathered through peripheral vision via covert attention (Rosenholtz, 2016; Wolfe & Whitney, 2015). For the purposes of this study, the term attention will refer exclusively to overt attention, understood as the selective focus on an object or spatial area achieved by directing the gaze toward it (Posner, 1980). Attention, defined as the ability to sustain a cognitive process to resolve a demand despite other distractors (Rueda et al., 2023; Shiferaw et al., 2019), can be influenced by bottom-up factors, such as colour, brightness, texture, and other salient features, or by goal-oriented and executive functions known as top-down factors (Baluch et al., 2011; Chica et al., 2013; Connor & Egeth, 2004; Katsuki & Constantinidis, 2014). Specifically, in art perception, these mechanisms interact together to direct attention towards the most salient features while also aligning with the creators intended focal points (Beelders & Bergh, 2020). Therefore, attention works as a selective filter, guiding visual focus to relevant information. Attention likely underwent a remarkable enhancement and specialization in the human genus, and as far as we can infer by changes in technological complexity, social organization, and fronto-parietal neuroanatomy (Bruner & Colom, 2022). In this sense, focused and executive attentional mechanisms, namely, the ability to control attention to ongoing cognitive processes, avoiding distractions and targeting to achieve a specific goal (Engle et al., 1999; Geva et al., 2013), contributed to evolve an intentional, sustained, and aware attentional process (Bruner, 2024). This attentional network largely relies on visual processing, which is the primary source of information in primates (Atkinson, 2008; Matsuno & Fujita, 2009). The selection and processing of relevant information is a prediction system that involves the management of external salience factors (bottom-up) and internal sources of information (top-down). According to active inference models of cognition, the brain continuously generates predictions to minimise errors between perceived information and previous experiences, enabled by actions such as eye movements (Clark, 2023; Constant et al., 2021; Friston et al., 2009). In addition to the observation of salient locations, saccades optimise selective visual sampling 123 J Cult Cogn Sci through continuous adjustments, with fixations directed toward the most significant areas of a scene (Shiferaw et al., 2019). This process as a whole allows for a comprehensive understanding of what is being observed. Palaeolithic art probably encodes messages and discourses that influence individual responses. Interpretation of these messages is conditioned by prior experience in visual comprehension, through a proper understanding of the relationship between symbol and meaning, and the perception/recognition of the symbols themselves (Pinna & Reeves, 2009). The composition, defined as the combination of elements within a painting, significantly influences visual behaviour, even if its meaning is not fully understood (Beelders & Bergh, 2020; Sancarlo et al., 2020). As mentioned, visual behaviour can be analysed by eye-tracking technique. The application of this method has a quite longstanding tradition within the field of art studies (e.g., Bailey-Ross et al., 2019; Buswell, 1935; Massaro et al., 2012; Molnar & Day, 1981; Rosenberg & Klein, 2015; Savazzi et al., 2014; Serino & Villani, 2015; Yarbus, 1967) and have also been applied in archaeological research (Criado-Boado et al., 2019,2023; Silva-Gago & Bruner, 2023; Silva-Gago et al., 2021,2022), indicating that visual attention patterns differ according to both biological and cultural changes. Regarding Palaeolithic art, some studies have analysed the visibility and positioning of depictions from multiple approaches (Garate et al., 2023; Intxaurbe et al., 2022; Ochoa & Garcı ´a-Diez, 2018; Rivero et al., 2024; Wisher et al., 2023b,c) while others focused on exploring the role of fire illumination in its visualization (Medina-Alcaide et al., 2021; Needham et al., 2022). The spatial distribution of paintings across different artworks seems to be determined by the cave’s morphology, which naturally provides forms and shapes for who created the depictions (Wisher et al., 2023b), and in some cases, they are located in areas highly visible to a large audience (Garate et al., 2023; Intxaurbe et al., 2022; Ochoa, 2017). On the other hand, a computerbased program developed for facial expressions showed that the anatomical parts most frequently depicted in rock engravings are the same as those considered most salient in photographs of present-day animals (Meyering et al., 2020). However, none of these studies have quantitatively analysed the visual attention directed to cave paintings. Despite all the approaches developed in specific case studies using other methodologies, the distribution of attention within a panel or depiction has not been quantified in a broader number of caves. The present study is conducted within the framework of the XSCAPE Project on Material Minds (ERC—Synergy Grant). This interdisciplinary project investigates the relationship between the material world and human cognition, with a particular focus on eye-tracking methodologies applied to the analysis of material culture. In this case, we applied this technology in a wide selection of the most iconic caves with Palaeolithic art. The first study was exploratory and involved a wide range of figurative and abstract representations across entire panels, while the second experiment includes only single figures and aimed to assess whether and to what extent incomplete figures are recognised and observed as if they were fully represented. Our research work focused on the relationship and perceptual hierarchy existing between depictions made by past humans (graphic motifs) and natural forms of the cave (wall elements of geological origin such as fissures, reliefs, etc.); and secondly, on the analysis of the perceptual hierarchy of the different anatomical regions that articulate a zoomorphic form. We test the null hypotheses that (i) attention depends on saliency of the cave wall visual features, and (ii) the anatomic representation of the figure does not influence the patterns of visual exploration. Experiment 1 Material and methods Participants. An opportunistic sample of 12 participants (5 females and 7 males) recruited from social media took part in the study. All subjects had normal or corrected-to-normal vision and they were aged between 19 and 55 years old (42 ±11 years). Most of them lived in an urban environment (92%) and had a university education (83%). All gave informed consent for their participation in the study, which was approved by the ethics committee of the Spanish National Research Council (CSIC). Stimuli. Twenty-six photographs of depictions from 15 iconic Palaeolithic caves on the Cantabrian seaboard and France (El Castillo, Covalanas, Llonı ´n, Trescalabres, La Pileta, Altamira, Las Chimeneas, 123 J Cult Cogn Sci Covaciella, Ekain, La Garma, Las Monedas, La Pasiega, Sotarriza, Tito Bustillo and Chauvet) were used in the analysis (Fig. 1of supplementary material). Five photographs of real animals (bisons and horses) were used as control images. Some depictions were single-figure as opposed to multiple panels (3 with two motifs, 1 with three motifs and 1 with seven motifs). Likewise, animal motifs were the most frequent theme (a total of 18 including bisons, deers, horses, aurochs and goats), five pictures with signs (rectangular, quadrangular, sinuous, grille, couplet, radiate, grill and lines), and two anthropomorphic or indeterminable shape. The degree of anatomical completeness in the animal figures has also been considered (13 complete and 5 incomplete) and their chronology (when it has been possible to determine it with certain guarantees: 7 pre-Magdalenian -approx. between 40.000 and 18.000 BPand 9 Magdalenian - approx. between 18.000 and 12.000 BP-). The distinction between these two categories is well-established in the literature and reflects two chronological phases. In general, Pre-Magdalenian figures are primarily defined by outlined forms, minimal anatomical detail, occasional disproportions, and a simplified stylistic approach. In contrast, Magdalenian representations exhibit greater proportional accuracy and a more naturalistic style. Procedure. Participants were asked to freely visually explore a set of 26 images for 10 s each while resting their heads on a chin rest in front of a 2800 monitor. The distance between the participant and the screen was 115 cm. Eye movements were recorded using the eye tracker device Eye Link 1000 Plus (SR Research, Canada), from the Material Minds Lab (MML), a facility provided by the XSCAPE project in the Institute of Heritage Sciences (INCIPIT—CSIC). Pupil position was sampled at 500 Hz. Eye Link 1000 Plus is a desk-mounted eye tracker consisting of a camera that automatically focuses on the subject’s eyes, illuminated by an infrared system. Stimulus presentation was built by a Python code which randomize the images for each participant. All participants viewed the same set of pictures, which were standardized to same size and resolution (1920 91080, 300 dpi). Eye position was calibrated by the fixation of nine predefined dots, sequentially presented on the monitor. Each trial started with a drift correction to ensure that calibration was valid. The calibration was repeated during the experimental session whether the participant left the chin rest for any reason or whether the calibration was not correct according to drift point validation. Data analysis. First, the areas of greatest visual saliency and therefore likely to attract attention were calculated by graph-based visual saliency (GBVS) computation algorithm using Matlab software (Fig. 1a; also see Silva-Gago et al., 2022). The Fig. 1 An example of saliency computation map (a) and fixation map (b). While saliency map highlight in red some fissures of the cave and angled areas, fixation map indicate that the most observed points are the horses’ heads (also in red). Below, an example of AOIs design (c) of the same panel according to the anatomical regions represented (right) and estimated when missing (e.g. anterior region of left depiction) 123 J Cult Cogn Sci algorithm decomposed the original image into distinct visual feature channels, such as colour, luminance, and orientation, and then computed contrast values at each location within the image for each of these individual feature channels. Then, different areas of interest (AOIs) were defined based on these most identified salient locations (called saliency AOI), other elements in the cave such as fissures, calcite formations and other changes in the morphology of the support (called cave AOI), as well as the depiction. Hereafter, saliency features refers to the most salient areas computed (in red colour) by the GBVS algorithm. Furthermore, some AOIs were defined based on the main anatomical regions of the figures, namely head, anterior,body and posterior (Fig. 1c). Missing anatomical regions of incomplete depictions were not considered and were not included in any AOI. We computed the dwell time (in milliseconds) relative to the size of each AOI (in pixels) as a measure of fixation density over an area (Silva-Gago et al., 2021). This variable (relative dwell time, DT_REL) standardizes the amount of attention directed to each AOI regardless of their size. According to this methodological approach, fixation count correlates with dwell time (Silva-Gago et al., 2022), thus it has been decided to show only the results of the second. Fixation maps were generated by Eye Link Data Viewer to visualize the most observed areas (Fig. 1b). Mann–Whitney test, Dunn test and Kruskal– Wallis test were run between AOIs and depictions in order to test differences. Bonferroni correction was applied in all tests. Percentages of fixation time were computed according to the fixation time directed the AOIs defined in the stimuli, excluding saccade movements. All analysis were conducted using R 4.4.1 (R Core Team, 2024), the ggplot2 (Wickham, 2016) and ggstastsplot packages (Patil, 2021). Results and discussion Considering all the pictures, Fig. 2shows the distribution of DT_REL across each AOI, demonstrating significant variability. Many visually salient areas received no fixations, while the depiction attracted more attention than both the salient regions of the photograph and other cave features, such as fissures (p \0.001). The painting accounted for 86% of the total fixation time, in contrast to only 8% for prominent areas and 6% for cave features likely to attract visual attention. The algorithm identifies the brightest or angular areas as the most salient ones, sometimes also coinciding with cracks or ridges where several fractures converge. However, they have barely been observed in comparison to the depiction even when they were used as an element of the figure (Wisher et al., 2023b). This is consistent with previous studies showing that more sensory salient features, namely the areas considered salient by computation algorithms, receive less attention when there are other underlying purposes that are influencing attention (Criado-Boado et al., 2019; Silva-Gago et al., 2021) (see Fig. 3). In zoomorphic paintings, the head was the most frequently fixated anatomical region (p \0.001; Fig. 2a). Human attention is inherently drawn to faces and heads (e.g. Cerf et al., 2009; Gregory et al., 2015; Hietanen, 1999; Leppa ¨nen, 2016; Langton et al., 2000), due to their social and communicative significance. In the current study, this distribution of visual attraction was similar to the control images of real animals (Fig. 2b), with no statistically significant differences between the observation of real horses or bisons and the cave representations (p [0.05). The similarity of the observation of the animals and the paintings could be related to the fact that the figures on the cave walls actually depict the animals in the environment. They are representative images of materiality. Fig. 2 Distribution of Relative Dwell Time for each AOI: cave fissures, calcite formation and other morphological changes (blue), depiction (red), computed saliency areas (grey). Upper bars show significant differences (p \0.05, Bonferroni correction) 123 J Cult Cogn Sci Regarding the sequence of visual attention, the head accounted for 71% of the total fixation time, while the other anatomical regions accounted for 29%. Initial fixations were most often directed at the centre of the stimuli, typically corresponding with the body of the Fig. (58% of cases), followed by the head, which was the next most frequently fixated region (85% of cases). The head is one of the most discriminant or recognisable parts of an animal (Meyering et al., 2020) so most attention is directed there to facilitate understanding of what is being observed. When comparing incomplete and complete figures, the distribution of visual attention was similar, the head remained the most observed region, displaying significant differences from the other anatomical regions (p \0.001). Nonetheless, complete figures attracted more fixations on the body and incomplete depictions on hind legs (Fig. 4). Visual behaviour was also similar in different periods (Fig. 5). There were no differences between Magdalenian depictions and paintings from previous chronologies (p [0.05). The head was the main focus of attention, regardless of the degree of completeness and the chronological period. It is also noteworthy that the attention in the incomplete Magdalenian depictions was more equally distributed among the different anatomical regions, particularly the head and the body. In summary, the focus on the head region is preferential to other anatomical regions, whether the figures are complete or incomplete, and throughout the Palaeolithic art’s temporal sequence. Experiment 2 Material and methods Participants. An opportunistic sample of 29 participants recruited from social media took part in the study, including 19 females and 10 males (40 ±13.3 years old). Most of them lived in an urban environment (86%) and all had university studies. All subjects had a normal or corrected-to-normal vision and gave informed consent for their participation in the study, which was approved by the ethics committee of the Spanish National Research Council (CSIC). Stimuli. A different sample of twenty-six photographs of depictions from 20 iconic Palaeolithic caves on the Cantabrian seaboard and France (El Castillo, Covalanas, Llonı ´n, Trescalabres, La Pileta, Altamira, Las Chimeneas, Covaciella, Ekain, La Garma, Las Monedas, La Pasiega, Sotarriza, Tito Bustillo and Chauvet, Lascaux, Cougnac, Le Portel and Niaux) were used in the analysis (Fig. 2of supplementary material). Three drawings of Postpaleolithic art and three pictures of prehistoric vessels Fig. 3 Distribution of Relative Dwell Time for each AOI in depictions (a) and control images of real animals (b). Upper bars show significant differences (p \0.05, Bonferroni correction) 123 J Cult Cogn Sci were used as distractor images, and were excluded from analysis. All depictions were single-figure and showing animal motifs (bison, doe, deer, horse, auroch, goat and bear). The sample of animal figures included 13 complete and 13 incomplete depictions. Most of the them were faced to the right (18), black colour was predominant in 11 panels, and red in 9 depictions (6 pictures include more than one colour). When it was possible to determine the chronological adscription, 10 panels were pre-Magdalenian (approx. between 40.000 and 18.000 BP) and 13 paintings were Magdalenian (approx. between 18.000 and 12.000 BP). Procedure. Participants were asked to observe 32 pictures for 10 s each following the same experimental set up as in Experiment 1 above. Eye movements were recorded using Eye Link 1000 Plus (SR Research, Canada) sampling pupil position at 500 Hz. Stimuli were presented using a python code which randomize the images for each participant. All participants viewed the same set of pictures. Eye position was calibrated by the fixation of nine predefined dots and each trial started with a drift correction to ensure that calibration is valid. Data analysis. Analysis was similar to previous Experiment 1. Saliency maps were generated using GBVS algorithm and Matlab software. The same AOIs were described based on computed saliency regions, cave characteristics and the depiction, as well as the anatomical regions of the figure (head, anterior, body and posterior). A depiction is considered Fig. 4 Distribution of Relative Dwell Time for each AOI in complete (left) and incomplete (right) depictions with figurative motifs. Upper bars show significant differences (p \0.05, Bonferroni correction) Fig. 5 Distribution of Relative Dwell Time for each AOI in Premagdalenian and Magdalenian depictions. Panels with no clear chronological association were excluded from analysis 123 J Cult Cogn Sci incomplete when one of these anatomical regions is missing. The main difference compared to Experiment 1 lies in the estimation of the missing region in the incomplete depictions according to the size of the depicted animal. While in Experiment 1, the missing regions of incomplete depictions were not considered, in Experiment 2, that missing regions (for example, the posterior legs) were estimated in order to know if the attention was directed to that area even if it was not represented (Fig. 1c). The centre of the images was also considered as an AOI (Fig. 6). We calculated the fixation time relative to the size (DT_REL) as the main variable to examine the allocation of attention across the panels (Silva-Gago et al., 2021). Fixation maps were generated by Eye Link Data Viewer to visualize the most observed areas. Mann–Whitney test, Dunn test and Kruskal–Wallis test were run between AOIs and depictions in order to test differences. Bonferroni correction was applied in all tests. All analysis were conducted using R 4.4.1 (R Core Team, 2024), the ggplot2 (Wickham, 2016) and ggstastsplot packages (Patil, 2021). Results and discussion Considering all the figures, the depiction attracted significantly more attention than the visually computed salient areas of the panels and other cave Fig. 6 Example of saliency map (a), average fixation map (b) and AOI design: saliency AOI (blue), cave AOI (orange), depiction (yellow) and centre of the image (black). Saliency AOI was described according to the most salient areas calculated in the saliency map (the reddest locations). When one fixation belongs to more than one AOI, it was included in both AOIs due to the impossibility to discriminate 123 J Cult Cogn Sci features (p \0.001; Fig. 7). The distribution of attention across saliency areas exhibited the greatest variability. The painting itself accounted for 90% of the total fixation time, whereas areas considered prominent were observed for only 7%, and cave features likely to attract attention were observed for just 3%. Hence, the painting received most of fixation time regardless of whether it uses the cave’s topography or not, in contrast to suggested in previous studies when the painting was removed (Wisher et al., 2023c). Regarding the anatomical regions, the head received the highest proportion of fixations (p \0.0001; Fig. 8). After the initial fixation, which was predominantly directed at the centre of the image due to centre bias (Doran et al., 2009; Ioannidou et al., 2016; Le Meur & Liu, 2015; Tatler, 2007), and coinciding with the body of the depicted animal, the head was the next most frequently observed area, attracting 40% of fixations (Fig. 9). The anterior region accounted for 20% of fixation time, the body 18%, and the posterior region 17%, while the centre of the image received only 5%. As previous studies have shown (Cerf et al., 2009; Hietanen, 1999; Langton et al., 2000; Meyering et al., 2020), the head is the region most observed and the most suitable for identifying an animal depicted. When comparing the overall distribution of visual attention of complete and incomplete depictions, the results were similar, suggesting that the degree of completeness did not influence the observation of the painting and the recognition of the animal represented (Fig. 10). Likewise, the sequence of gaze direction was similar, with the head being the following fixated area after the first fixation directed to the centre of the picture. Furthermore, in a chronological comparison, the fixation pattern was similar, the head is the most observed region. However, significant differences were found in the fixation time on both extremities (p \0.001), while the body and head showed similar values. When examining by degree of completeness, complete Magdalenian figures were generally more observed, particularly in the head region, compared to pre-Magdalenian paintings. On the other hand, in incomplete figures, the head showed similar fixation times, but there were differences across the rest of the body, namely, incomplete Magdalenian figures received more attention on the posterior regions, whereas pre-Magdalenian figures were more frequently observed on the body. Furthermore, in Magdalenian period, there are more differences between complete and incomplete depictions than in previous chronologies (Fig. 11). This difference may be related to the greater degree of anatomical detail in the Fig. 7 Distribution of Relative Dwell Time for each AOI: cave fissures, calcite formation and other morphological changes (blue), depiction (red), computed saliency areas (grey). Upper bars show significant differences (p \0.05, Bonferroni correction) 123 J Cult Cogn Sci Archaeology, 52(2), 205–222. https://doi.org/10.1080/ 00438243.2020.1891964 Moberget, T., Gullesen, E. H., Andersson, S., Ivry, R. B., & Endestad, T. (2014). Generalized role for the cerebellum in encoding internal models: Evidence from semantic processing. Journal of Neuroscience, 34(8), 2871–2878. https://doi.org/10.1523/JNEUROSCI.2264-13.2014 Molnar, F., & Day, H. I. (1981). Advances in intrinsic motivation and aesthetics (pp. 385–413). About the Role of Visual Exploration in Aesthetics. Plenum Press. Mudrik, L., Faivre, N., & Koch, C. (2014). Information integration without awareness. Trends in Cognitive Sciences, 18(9), 488–496. https://doi.org/10.1016/j.tics.2014.04.009 Munar, E., Go ´mez-Puerto, G., Call, J., & Nadal, M. (2015). Common visual preference for curved contours in humans and great Apes. PLoS ONE, 10(11), Article e0141106. https://doi.org/10.1371/journal.pone.0141106 Needham, A., Wisher, I., Langley, A., Amy, M., & Little, A. (2022). Art by firelight? Using experimental and digital techniques to explore Magdalenian engraved plaquette use at Montastruc (France). PLoS ONE, 17(4), Article e0266146. https://doi.org/10.1371/journal.pone.0266146 Ochoa, B. 2017. Espacio gra ´fico, visibilidad y tra ´nsito cavernario. British Archaeological Reports International Series 2875. Ochoa, B., & Garcı ´a-Diez, M. (2018). The use of cave art through graphic space, visibility and cave transit: A new methodology. Journal of Anthropological Archaeology, 49, 129–145. https://doi.org/10.1016/j.jaa.2017.12.008 Ochoa, B., Vigiola-Ton ˜a, I., & Garcı ´a-Diez, M. (2019). Coming back to Ekain (Deba, Gipuzkoa): A new Upper Palaeolithic graphic ensemble in the Erdibide Passage. Munibe Antropologia Arkeologia, 70, 65–71. https://doi.org/10. 21630/maa.2019.70.10 Onians, J. (2024). Understanding rock art: What neuroscience Can Add. In O. M. Abadı ´a, M. W. Conkey, & J. McDonald (Eds.), Deep-time images in the age of globalization: Rock art in the 21st Century (pp. 181–192). Springer International Publishing. https://doi.org/10.1007/978-3-03154638-9_12 Patil, I. (2021). Visualizations with statistical details: The ‘‘ggstatsplot’’ approach. Journal of Open Source Software, 6(61), 3167. https://doi.org/10.21105/joss.03167 Persike, M., & Meinhardt, G. (2016). Contour integration with corners. Vision Research, 127, 132–140. https://doi.org/10. 1016/j.visres.2016.07.010 Pinna, B., & Reeves, A. (2009). From perception to art: How vision creates meanings. Spatial vision.https://doi.org/10. 1163/156856809788313147 Posner, M. I. (1980). Orienting of attention. The Quarterly Journal of Experimental Psychology, 32(1), 3–25. https:// doi.org/10.1080/00335558008248231 R Core Team (2021). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/. Accessed 17 July 2024. Rivero, O., Beato, M. S., Alvarez-Martinez, A., Garcı ´a-Bustos, M., Suarez, M., Mateo-Pellitero, A. M., Eseverri, J., & Eguilleor-Carmona, X. (2024). Experimental insights into cognition, motor skills, and artistic expertise in Paleolithic art. Scientific Reports, 14(1), 18029. https://doi.org/10. 1038/s41598-024-68861-2 Robert, E. (2017). The role of the cave in the expression of prehistoric societies. Quaternary International, 432, 59–65. https://doi.org/10.1016/j.quaint.2015.11.083 Rolls, E. T. (2000). Functions of the primate temporal lobe cortical visual areas in invariant visual object and face recognition. Neuron, 27(2), 205–218. https://doi.org/10. 1016/S0896-6273(00)00030-1 Rosenberg, R., & Klein, C. (2015). The moving eye of the beholder. Eye-tracking and the perception of paintings. In J. P. Huston, M. Nadal, F. Mora, L. F. Agnati, & C. J. CelaConde (Eds.), Art, aesthetics and the brain (pp. 79–108). Oxford University Press. Rosenholtz, R. (2016). Capabilities and limitations of peripheral vision. Annual Review of Vision Science, 2(1), 437–457. https://doi.org/10.1146/annurev-vision-082114-035733 Rueda, M. R., Moyano, S., & Rico-Pico ´, J. (2023). Attention: The grounds of self-regulated cognition. Wires Cognitive Science, 14(1), Article e1582. Ruiz-Redondo, A. (2016). Le comportement symbolique des derniers chasseurs cueilleurs pale ´olithiques: Regard sur l’art rupestre du Magdale ´nien cantabrique. L’anthropologie, 120(5), 568–587. https://doi.org/10.1016/j.anthro. 2016.10.002 Sakamoto, T., Pettitt, P., & Ontan ˜on-Peredo, R. (2020). Upper Palaeolithic Installation Art: Topography, distortion, animation and participation in the production and experience of Cantabrian cave art. Cambridge Archaeological Journal, 30(4), 665–688. https://doi.org/10.1017/ S0959774320000153 Sancarlo, R., Dare, Z., Arato, J., & Rosenberg, R. (2020). Does pictorial composition guide the eye? Investigating four centuries of last supper pictures. Journal of Eye Movement Research.https://doi.org/10.16910/jemr.13.2.7 Savazzi, F., Massaro, D., Di, C., Gallese, V., Gilli, G., & Marchetti, A. (2014). Exploring responses to art in adolescence: A behavioral and eye-tracking study. PLOSONE. https://doi.org/10.1371/journal.pone.0102888 Shevelev, I. A., Kamenkovich, V. M., & Sharaev, G. A. (2003). The role of lines and corners of geometric figures in recognition performance. Acta Neurobiologiae Experimentalis, 63(4), 361–368. Shiferaw, B., Downey, L., & Crewther, D. (2019). A review of gaze entropy as a measure of visual scanning efficiency. Neuroscience & Biobehavioral Reviews, 96, 353–366. https://doi.org/10.1016/j.neubiorev.2018.12.007 Sigman, M., Cecchi, G. A., Gilbert, C. D., & Magnasco, M. O. (2001). On a common circle: Natural scenes and Gestalt rules. Proceedings of the National Academy of Sciences of the United States of America, 98(4), 1935–1940. https:// doi.org/10.1073/pnas.98.4.1935 Silva-Gago, M. and Bruner, E. (2023). Cognitive Archaeology, attention, and visual behavior. In Bruner (ed). Cognitive Archaeology, Body Cognition, and the Evolution of Visuospatial Perception, pp. 213–239. https://doi.org/10. 1016/b978-0-323-99193-3.00013-1. Silva-Gago, M., Fedato, A., Hodgson, T., Terradillos-Bernal, M., Alonso-Alcalde, R., & Bruner, E. (2021). Visual attention reveals affordances during Lower Palaeolithic 123 J Cult Cogn Sci stone tool exploration. Archaeol. Anthropol. Sci., 139(13), 1e11. Silva-Gago, M., Ioannidou, F., Fedato, A., Hodgson, T., & Bruner, E. (2022). Visual attention and cognitive archaeology: an eye-tracking study of palaeolithic stone tools. Perception, 51(1), e324. Spillmann, L. (2006). Chapter 5 From perceptive fields to Gestalt. In S. Martinez-Conde, S. L. Macknik, L. M. Martinez, J.-M. Alonso, & P. U. Tse (Eds.), Progress in brain research (Vol. 155, pp. 67–92). Elsevier. https://doi. org/10.1016/S0079-6123(06)55005-8 Tatler, B. W. (2007). The central fixation bias in scene viewing: Selecting an optimal viewing position independently of motor biases and image feature distributions. Journal of Vision, 7(14), 1–17. https://doi.org/10.1167/7.14.4 Villani, D., Morganti, F., Cipresso, P., Ruggi, S., Riva, G., & Gilli, G. (2015). Visual exploration patterns of human figures in action: An eye tracker study with art paintings. Frontiers in Psychology, 6, 1636. https://doi.org/10.3389/ fpsyg.2015.01636 Wickham H (2016). ggplot2: Elegant graphics for data analysis. Springer-Verlag New York. ISBN 978-3-319-24277-4, https://ggplot2.tidyverse.org. Accessed 8 Oct 2024. Webster, M. A., & MacLeod, D. I. (2011). Visual adaptation and face perception. Philosophical Transactions of the Royal Society B: Biological Sciences, 366(1571), 1702–1725. https://doi.org/10.1098/rstb.2010.0360. Wertheimer, M. (1938). Laws of organization in perceptual forms. London: Harcourt, Brace & Jovanovitch Wisher, I., Pagnotta, M., Palacio-Pe ´rez, E., Fusaroli, R., Garate, D., Hodgson, D., Matthews, J., Mendoza-Straffon, L., Ochoa, B., Riede, F., & Tyle ´n, K. (2023a). Beyond the image: Interdisciplinary and contextual approaches to understanding symbolic cognition in Paleolithic parietal art. Evolutionary Anthropology: Issues, News, and Reviews, 32(5), 256–259. https://doi.org/10.1002/evan. 21996 Wisher, I., Pettitt, P., & Kentridge, R. (2023). Conversations with caves: The role of Pareidolia in the Upper Palaeolithic Figurative Art of Las Monedas and La Pasiega (Cantabria Spain). Cambridge Archaeological Journal.https://doi. org/10.1017/S0959774323000288 Wisher, I., Pettitt, P., & Kentridge, R. (2023). The deep past in the virtual present: Developing an interdisciplinary approach towards understanding the psychological foundations of palaeolithic cave art. Scientific Reports.https:// doi.org/10.1038/s41598-023-46320-8 Wolfe, B. A., & Whitney, D. (2015). Saccadic remapping of object-selective information. Attention, Perception & Psychophysics, 77(7), 2260–2269. https://doi.org/10.3758/ s13414-015-0944-z Yan, Y., Ren, J., Sun, G., Zhao, H., Han, J., Li, X., Marshall, S., & Zhan, J. (2018). Unsupervised image saliency detection with Gestalt-laws guided optimization and visual attention based refinement. Pattern Recognition, 79, 65–78. https:// doi.org/10.1016/j.patcog.2018.02.004 Yarbus, A. L. (1967). Eye Movements and Vision. Springer. Yu, G., Katz, L. N., Quaia, C., Messinger, A., & Krauzlis, R. J. (2024). Short-latency preference for faces in primate superior colliculus depends on visual cortex. Neuron, 112(16), 2814-2822.e4. https://doi.org/10.1016/j.neuron. 2024.06.005 Zupan, Z., & Gvozdenovic ´, V. (2023). Visual search of illusory contours: The role of illusory contour clarity. Attention, Perception, & Psychophysics, 85(2), 578–584. https://doi. org/10.3758/s13414-022-02644-7 Publisher’s Note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. 123 J Cult Cogn Sci