Introduction

In the discovery and authentication of Paleolithic art1,2, one of the main features that has attracted the attention of early researchers to this graphic production was the high degree of technical perfection of some of its representations and the cognitive development involved, attributed to Anatomically Modern Humans. Visual art is considered since then one of the hallmarks of humanity, linked to modern human behavior around the world3,4,5. Paleolithic graphic productions, starting 51,200 years ago, represent some of the earliest examples of creativity and cognitive complexity attributed to Anatomically Modern Humans6. However, although Paleolithic art constitutes the first and one of the clearest examples of realistic drawing generated by Homo sapiens, there is a notable scarcity of studies that have used Paleolithic art as a basis for understanding the cognitive processes involved in the perception and execution of this representational art. In the context of Paleolithic art, the term “art” is used to describe a set of technical and stylistic skills that are characteristic of a specific savoir-faire in each region and period. It is important to note that this term does not imply any aesthetic connotations. For its part, the term “art” and "graphic production" are used to describe all non-functional representations created by the first Anatomically Modern Humans.

Unravelling how the first artists engaged different cognitive processes (e.g., visuospatial ability or perceptual and attentional processes) and motor skills to achieve advanced drawing performance is fundamental for understanding artistic expertise, a concept defined in psychological research as having experience and/or skill in visual art production7, and the cognitive development of our species. Since drawing and artmaking are viewed as a window into human thought and action8,9, the graphic expressions of our first ancestors can be considered a means to understand how early representatives of our species acquired the spatial, perceptual, and motor control skills needed to create Paleolithic motifs10.

In recent years, notable advances have been made in the field of cognitive archaeology, which seeks to provide answers about the cognitive abilities inherent to our species, particularly in the field of lithic technology11,12,13,14 and, more recently, about the specific cognitive abilities associated with the symbolic behavior of Homo sapiens15,16. These studies employ experimental reference protocols that assume that there are sufficient neurological bases to establish connections between Paleolithic individuals and current humans, despite genetic and cultural variations16,17.

At the same time, Psychology has made significant contributions to the understanding of human cognitive processes. Through empirical and experimental research, psychologists have studied how our cognitive functions are influenced by biological factors, such as genetics, as well as social and cultural factors, among others. Specifically, some studies have explored the cognitive processes associated with the creation of visual art. These investigations have focused on understanding perceptual and learning processes in artmaking, with particular emphasis on discriminating cognitive abilities between artists and non-artists, measured through psychometric approaches10,18,19,20,21,22,23,24,25.

When studying the conditions necessary to achieve mastery in a specific ___domain, it is crucial to quantify the level of expertise in that ___domain. Traditionally, in expertise research, fields such as sports have been extensively investigated26,27, partly because they represent measurable domains. However, exploring the factors leading to expertise in the visual arts is particularly challenging, primarily due to the non-homogeneous nature of the group of artists in terms of subject matter or artistic medium chosen in their works28. Despite the methodological challenges associated with quantifying factors in artistic production, some research efforts have endeavored to elucidate the relevant cognitive processes in the realm of artistic expertise20,23,25,29,30,31,32,33,34.

In the few available studies, artistic expertise has been found to be associated with an enhanced capacity to attend to, manipulate, or process specific aspects of visual information more efficiently7,23,28,31,35. This reveals that artists have an advantage over non-artists in various perceptual, attentional, and spatial skills. For instance, artists have exhibited superior performance in object identification in unfocused photographs and in detecting embedded patterns in more complex images20. Furthermore, other skills important to the visual arts include visual memory36, flexibility in shifting between global and local attention22, the ability to integrate local details into global representations of objects35, and the ability to generate and transform mental images18. However, it is noteworthy that while these findings emphasize the existence of specific cognitive skills in artists, the scarce results have not always been consistent, with studies failing to find the expected differences in favor of artists7,22,23,30,37,38. In conclusion, given the paucity of research on visual arts expertise and the inconsistency of the findings, the relationship between cognitive abilities and artistic expertise remains uncertain in current research.

Additionally, ongoing debates persist regarding the basis of expert knowledge and the reasons why artists produce realistic depictions. In particular, the role of innate talent7,39,40, the effects of artistic media41, and the integration between perception and motor skills in drawing abilities7 continue to be discussed. The difficulty of reaching firm conclusions in these areas is in part due to the nature of the cognitive assessments used in these investigations. These assessments have often been very general, requiring the engagement of diverse abilities. However, a more appropriate approach would be to advocate the use of cognitive tests tailored to specific cognitive abilities. This has already proven relevant for visuospatial cognitive processing and for psychomotor skills involving motor movements, precision, coordination, and strength42. Previous research focused on addressing the motor skills necessary to achieve superior drawing performance has used gestural analysis of Paleolithic engraving production43,44,45. Rivero and Garate45 conducted a comparative analysis between individuals with no prior experience in engraving and experts with over a decade of expertise in Paleolithic engraving. Their findings revealed the existence of gestural parameters that influence the interaction with the tool and the support, enabling expert engravers to create visible and technically accurate motifs. The results of this experimental study corroborate the evidence of artistic learning highlighted by the analysis of archaeological material44,46,47. This evidence is based on the identification of microscopic indices, called stigmas, that reveal inexperience in the handling of the tool.

Artistic and technical practices encompass pedagogical, imitative, and other forms of social learning, facilitating the transmission of information that shapes cultural traditions between individuals and across generations. The development of artistic skills occurs within "communities of practice"48,49, wherein the collective knowledge of the group's members is utilized to advance the group's collective expertise. The study of the chaînes opératoires involved in the production of graphic motifs has enabled researchers to overcome the constraints faced by research on the symbolic behavior of Upper Paleolithic societies. Technical analysis establishes a link between the art and the wider archaeological record, thereby endowing Paleolithic art with social and cultural meaning50. The identification of learning processes yields information regarding individual cognition, cultural structure, intergroup relations, changes over time and space, and the role of art within Paleolithic societies44,46. However, there is still a significant gap in our understanding of how artistic skills developed in the Anatomically Modern Humans and the cognitive abilities involved in the creation of this artwork.

Pioneering a coherent interdisciplinary approach, the present research integrates experimental archaeology, cognitive psychology, and biomechanics of gesture to identify the cognitive abilities and characterize the psychomotor aspects involved in the creation of Paleolithic art motifs by comparing individuals with artistic expertise (referred to as “experts” hereafter) and individuals without artistic expertise (referred to as “non-experts” hereafter). Carefully selected materials and cognitive tests, based on previous research findings, have been used to answer the following research question: What cognitive abilities and motor skills are involved in the creation of art and engravings? In Study 1, we investigated the cognitive abilities of experts and non-experts in visual arts, whereas in Study 2, our interest was specifically focused on analyzing the motor skills exhibited by experts and non-experts during drawing and engraving with Paleolithic techniques. By addressing this question, we contributed to a deeper understanding of the cognitive and motor processes underlying Paleolithic artistic production.

Study 1

Study 1 explored the cognitive abilities involved in the creation of visual art. Particularly, we focused on some cognitive processes traditionally associated with drawing and artmaking: spatial ability and memory.

On the one hand, although recent emphasis has been placed on the relevance of spatial ability for studying cognitive development in our species42, it is noteworthy that spatial ability is one of the less-explored cognitive abilities in cognitive archaeology. Spatial ability refers to the “skill in representing, transforming, generating, and recalling symbolic, nonlinguistic information”51p. 1482. Despite the lack of consensus on the factors underlying spatial ability, previous research has identified three factors: spatial visualization, mental rotation, and perceptual speed52,53,54,55. In previous studies, the spatial ability of visual art experts was typically limited to either spatial visualization or mental rotation skills18,29,30,56,57,58,59, showing inconsistent results. In Study 1, we independently analyzed spatial visualization, mental rotation, and perceptual speed to gain an understanding of the spatial abilities that contribute to artistic expertise.

On the other hand, some studies have suggested that differences in spatial ability may be due to differences in working memory53. As working memory is responsible for temporarily holding information to actively process it for a specific task, it could play a crucial role in tasks involving mental manipulation and representation of objects in space, making it indispensable for the creation of art. Moreover, it is essential to recall visual details, object shapes, and their spatial arrangement while painting or drawing32. Therefore, both short-term memory and working memory are engaged in such tasks. Given the importance of visual memory in art production36, it would be reasonable to assume that memory is a factor that may help explain cognitive differences between experts and non-experts in visual arts. Thus, in Study 1 we included tests assessing both working memory and short-term memory.

In summary, given the scarcity of research and the inconsistency of the results regarding cognitive advantages associated with artistic expertise, in this study, for the first time in the literature, we jointly analyzed three factors of spatial ability (spatial visualization, mental rotation, and perceptual speed) and memory capacity (short-term memory and working memory) to clarify the differences between experts and non-experts in visual art.

Method

Participants

A total of 100 undergraduate students (Mage = 20.68, SDage = 1.50; 79% of women) agreed to participate in this study and signed a consent form. The participants were divided into two groups based on whether they had artistic expertise or not: experts versus non-experts. The group of experts comprised 50 students of Fine Arts from the University of Salamanca (11 men and 39 women), who had varied prior formal training in drawing or painting. The group of non-experts included 50 Psychology students (10 men and 40 women) from the same university, who had no prior formal training in drawing or painting and, therefore, had no artistic expertise. The study was approved by the Ethics Committee of the University of Salamanca and was conducted in accordance with the Declaration of Helsinki.

Materials

To assess spatial ability and memory, the following psychometric tests were used.

Firstly, we assessed spatial ability through spatial visualization, mental rotation, and perceptual speed tasks. Spatial visualization was measured by the Space Relations (SR) subtest of the Differential Aptitude Test (DAT-5)60. The DAT Space Relations test is a mental folding test consisting of 50 items, each composed of an unfolded figure and four folded alternatives. Participants were asked to identify the folded figure that matched the unfolded figure on the left. The score is the total number of correct responses. To measure mental rotation, we used the Spatial subtest of the Thurstone’s Primary Mental Abilities (PMA-S)61. The PMA-S comprises 20 items, each including a flat model figure and six alternatives that must be evaluated against it. Some alternatives were simply rotated versions of the model figure, while the remaining ones were mirror images. The task was to select only the rotated figures (several alternatives could be correct for each item). The score is the total number of correct responses (i.e., appropriately selected figures) minus the total number of incorrect responses (i.e., inappropriately selected figures). Finally, perceptual speed was assessed using the Clerical Speed and Accuracy (CSA) subtest of the Differential Aptitude Test (DAT-5)60. This task involved finding letter and/or number combinations in strings of random letters and/or numbers. The test consisted of two parts, each with 100 items, and lasted 3 min. Only the second part was considered, with a maximum score of 100 points.

Secondly, two types of memory were measured, short-term memory (STM) and working memory (WM), using both verbal and visuospatial information. To assess verbal STM and WM, we used the Forward and Backward Digit Span subtests of the Wechsler Adult Intelligence Scale (WAIS-IV)62, respectively. In these subtests, participants repeated increasingly longer strings of numbers forward or backwards, with the forward condition always presented first. The length of the number strings ranged from two to nine for the forward condition and from two to eight for the backward condition. Two different sequences of numbers were presented for each length, making a total of 16 and 14 trials, respectively. The test ended when participants failed to recall both trials for a given length. To assess visuospatial STM and WM, we employed the Forward and Backward Spatial Span subtests of the Wechsler Memory Scale (WMS-III)63, respectively. In this task, the examiner touched a sequence of blocks (one per second) in different locations that were mounted on a 10-block board, and participants were asked to reproduce the sequence either in the same (forward) or reverse (backward) order. The number of blocks to be touched in each sequence increased from two to nine, with two block sequences for each length. The test ended when participants failed in both trials for a given length.

Procedure

All participants were tested individually, and the session lasted approximately one hour. Before starting the study, participants were presented with an informed consent form to read and sign, followed by a demographic questionnaire collecting data on age, sex, handedness, hours of sleep, potential vision corrections (i.e., contact lenses or glasses), and any artistic expertise (i.e., formal training in painting, drawing, or engraving). Next, the Forward and Backward Digit Span subtests (FDS and BDS, respectively) were administered, followed by the SR and the CSA subtests of the DAT. Subsequently, the PMA-S was completed, followed by the application of the Forward and Backward Spatial Span subtests (FSS and BSS, respectively). Once the tasks were completed, participants were thanked for their participation, any questions they had were answered, and the research purpose was briefly explained to them.

Data analyses

Data analyses were conducted using the JASP software64. Alpha was set at 0.05 for all analyses. Effect sizes were reported using Cohen’s d, with 95% confidence intervals (CIs) provided. Independent samples t-tests were performed to examine differences between experts and non-experts in the assessed cognitive abilities. Additionally, Pearson correlation analyses were conducted to investigate potential relationships between different cognitive abilities.

Results and discussion

Table 1 shows the descriptive statistics of the cognitive abilities measured in Study 1 and the results of the independent samples t-tests used to analyze the differences between experts and non-experts in memory (verbal and spatial STM and WM) and spatial ability (spatial visualization, mental rotation, and perceptual speed).

Table 1 Mean (SD) differences in the cognitive abilities between experts and non-experts in Study 1.
Full size table

No significant differences between experts and non-experts were found in the Forward Digit Span subtest, t(98) = 1.68, p = 0.095, Backward Digit Span subtest, t(98) = 0.37, p = 0.710, Forward Spatial Span subtest, t(98) = 0.60, p = 0.549, nor in the Backward Spatial Span subtest, t(98) = 0.79, p = 0.432. In other words, there were no significant differences between experts and non-experts in any of the memory tests, whether involving verbal or spatial content, neither in short-term memory nor in working memory.

Regarding the assessment of spatial abilities, there were no significant differences between the groups in CSA, a test measuring perceptual speed, t(98) = 0.57, p = 0.570. However, significant differences were found in spatial visualization between experts (M = 34.04, SD = 8.14) and non-experts (M = 28.20, SD = 7.16), as revealed by the Space Relations subtest of the DAT, t(98) = 3.81, p < 0.001, Cohen’s d = 0.76, 95% CI [0.35, 1.16]. Furthermore, significant differences were also found between groups in PMA-S, a mental rotation test, t(98) = 2.67, p = 0.009, Cohen’s d = 0.53, 95% CI [0.13, 0.93], where experts (M = 29.14, SD = 9.03) scored higher than non-experts (M = 23.70, SD = 11.21). In other words, individuals with artistic expertise scored higher in spatial ability, both in spatial visualization and mental rotation, than those without such expertise. However, similar to memory, there were no significant differences between experts and non-experts in perceptual speed.

Finally, to gather more information regarding the potential relationship between the cognitive abilities, a correlational analysis was conducted (see Table 2).

Table 2 Pearson correlation matrix between the cognitive abilities in Study 1.
Full size table

Positive correlations were observed between the Forward and Backward Digit Span subtests, r = 0.513, p < 0.001, and between the Forward and Backward Spatial Span subtests, r = 0.367, p < 0.001. In other words, our results showed that there was a relationship between short-term memory and working memory, for both verbal and visuospatial information. In line with previous research, these results indicated that higher scores in working memory were associated with higher scores in short-term memory65,66.

More relevant to our study, results revealed a significant positive correlation between SR and PMA-S, r = 0.345, p < 0.001, suggesting that the higher the spatial visualization ability, the higher the mental rotation ability. However, neither of them correlated with the CSA, a perceptual speed test (see Table 2). Additionally, both SR and PMA-S, but not the CSA, showed significant positive correlations with Backward Spatial Span, r = 0.266, p = 0.007; r = 0.289, p = 0.004, respectively. This implies that higher scores in spatial working memory were associated with higher scores in mental rotation and spatial visualization, but not in perceptual speed. Furthermore, only PMA-S (a mental rotation test) positively correlated with Forward Spatial Span, r = 0.290; p = 0.003, suggesting that high scores in short-term memory for spatial information are associated with high scores in mental rotation.

Study 2

In Paleolithic art, the technical analysis of artistic production has revealed the presence of learning processes during the European Upper Paleolithic44. These processes, identified through experimental protocols and the analysis of the archaeological record, allow us to infer the existence of artists and distinguish between experts and non-experts in the Upper Paleolithic, as well as the motor aspects related to Paleolithic graphic production45. However, no biomechanical analysis of gesture has been conducted to date in the study of Paleolithic graphic production, unlike other technical fields such as lithic reduction12,67. Thus, the purpose of Study 2 was to conduct a biomechanical study of hand movements in experts and non-experts in visual art when performing various graphic production tasks with both current techniques and prehistoric instruments. This study aimed to obtain data on motor skills associated with expertise in graphic production and to further expand upon the findings from Study 1. Both psychometric tests of cognitive abilities (Study 1) and motor analyses (Study 2) were carried out on the same population to elucidate whether artistic expertise led to cognitive and psychomotor differences compared to non-expert participants.

Method

Participants

In this study, a total of 33 undergraduate students (Mage = 21.46, SDage = 2.50; 63.6% of women) participated voluntarily. They all agreed to take part and signed a consent form. As in Study 1, participants were divided into two groups based on their artistic expertise: experts versus non-experts. The group of experts included 12 Fine Arts students (2 men and 10 women) from the University of Salamanca. The group of non-experts included 21 Psychology students (10 men and 12 women) from the same university who did not have any prior artistic expertise. The study was approved by the Ethics Committee of the University of Salamanca and was conducted in accordance with the Declaration of Helsinki.

Materials

To assess the biomechanics of gesture and the drawing and engraving skills related to Paleolithic art production, the following tasks were performed. Following previous research68, participants were asked to make three horse figures, the most represented animal in Paleolithic art69.

Task 1 required participants to complete a drawing from memory task, where they were asked to draw a horse from memory on paper. This task aimed to assess the ability of both experts and non-experts to generate a realistic image from long-term memory.

Task 2 was also a drawing from memory task, but in this case, participants were first shown a tracing of a horse figure depicted in Paleolithic art from the site of Les Trois Frères70, known for its large quantity of anatomical detail, and then asked to draw that particular horse from memory on paper. This task primarily assessed participants' visual short-term memory for the formal characteristics of the figures and their fine motor skills involved in drawing them.

Task 3 involved reproducing the same image used in Task 2 from memory, but this time using a media and tool representative of Paleolithic art. To do so, participants employed a dihedral burin, an archaeological tool associated with engraving artistic motifs71. The motifs were made on a schist slab, a material suitable for engraving and used in Paleolithic art72. This task aimed to discriminate the specific biomechanical characteristics of engraving from previous drawing from memory tasks and to assess the abilities of both experts and non-experts to produce the motifs using Paleolithic tools and procedures.

During the execution of Tasks 1, 2 and 3, participants' movements were monitored with Rokoko Smartgloves, from which the coordinates of the phalanges' positions were obtained. The movements were also recorded on video with a Nikon D850 camera (see Figure S1 in Supplementary Information).

Procedure

The protocol was developed in the Prehistoric Technology Laboratory of the University of Salamanca. Participants completed the tasks individually, and the duration of the study was approximately 30 min. Before starting, participants were presented with an informed consent form to read and sign. Afterwards, participants completed a demographic questionnaire collecting data on age, sex, height, level of education, and artistic expertise (i.e., formal training in painting, drawing, or engraving). Next, the Edinburgh Handedness Inventory73 was applied to assess the participants' laterality. Finally, forearm measurement was taken from each participant, and the Rokoko Smartgloves were calibrated using the reference posture (see Figure S2 in Supplementary Information).

The protocol began with drawing a horse from memory in pencil on paper for one minute (Task 1). Afterwards, a tracing of a horse figure from the cave of Les Trois Frères appeared on a computer screen for two minutes, with a size of 34 × 23 cm and at a distance of 2 m (Task 2). Figure orientation varied according to the laterality of the participants, with the motif oriented to the left for right-handed participants and to the right for left-handed participants, following previous research that established a relationship between manual dominance and the sequence followed in graphic representation74. After the image disappeared, participants had to draw the horse from memory on a sheet of paper with the same pencil as in Task 1. Participants were given a maximum of two minutes to complete this task. Finally, participants reproduced from memory the horse previously shown on the screen using a burin to engrave the figure on a shale plate (Task 3). The maximum time provided for this task was 10 min (see Fig. 1), as previous experimental studies on Paleolithic engraving45 have shown that the time available for the task did not influence the result. Participants were instructed to copy the stimulus as accurately and efficiently as they could.

Figure 1
figure 1

(a) Task 2: drawing from memory a representation of Paleolithic art in pencil on paper. (b) Task 3: engraving from memory a representation of Paleolithic art on a lithic media.

Full size image

Data analyses

Biomechanics of gesture

The analyses of the biomechanics of gesture aimed to differentiate grip patterns according to the tools used, as well as different movements depending on the task performed. To this end, an executable (BHV2ANGLES) was generated to process the data provided by the Rokoko Smartgloves to obtain the anatomical angles of flexion, extension, abduction, and adduction of the fingers of the hand: Flexion/Extension and deviation of the wrist (WRIST_F and WRIST_A), Flexion/Extension of the proximal phalanges (PIP_F), Flexion/Extension of the metacarpophalangeal (MCP_F), Relative abduction/adduction of each finger (MCP_A), and Flexion/Extension and Abduction/Adduction of the carpometacarpal of the thumb (CMC) (see Figure S3 in Supplementary Information).

Based on the data generated by the BHV2ANGLES software, we analyzed the data through multivariate statistics to compare gesture sequences between experts and non-experts, as well as between tasks. The data obtained were processed using the open-access R Statistical Software (V.4.2.3)75. The analysis required the FactoMineR: Multivariate Exploratory Data Analysis and Data Mining package76. Principal Component Analysis (PCA) and Cluster Analysis (Ward's method and Euclidean distance) were performed to identify the formation of groups based on the similarity in the use of certain hand positions to a greater or lesser extent.

Analyses of realistic image generation, and drawing/engraving of Paleolithic motifs

For the analyses of the drawn figures, Tasks 1 and 2 were scanned. For its part, for the analyses of the engraved figures, digital tracings of the Task 3 figures were made. The generated .jpg files were converted to a .dxf CAD data file format using an online converter. This transformation allowed vectorization, enabling automated measurements through AutoCAD software. In Task 3, close-range photogrammetry was applied to the engraved slabs produced by the participants77. The images were processed using Agisoft Metashape software to generate highly detailed 3D models with flash-type grazing lighting78. These models were then scaled to obtain the topography of both the surface of the media and the engraving. Subsequently, formal and technical analyses were conducted to characterize various aspects of the representations produced by the participants.

In formal analyses, firstly, we analyzed the similarity between the figure used as a model and those reproduced by the participants. For these analyses, we considered the size of the parts of the drawn/engraved figures with respect to the size of the original model. To evaluate similarity, a mathematical approach was used for the first time in this field. This approach involved measuring Euclidean distances and normalizing them to obtain a quantitative “similarity index” (see Fig. 2, and Table S1 in Supplementary materials). In Task 1, the similarity index was calculated using measurements taken from a real horse and comparing them with the drawing produced by the participants. For Tasks 2 and 3, reference measurements were taken from the representation of Les Trois-Frères to compare them with the measurements of the motifs produced by the participants.

Figure 2
figure 2

Measurements taken on the sample representation and on an example of the representations produced in Study 2 to calculate the Euclidean distance.

Full size image

This process started with the selection of key measurements to obtain specific values in the image to be replicated. A total of 19 measurements were chosen, enough to characterize the size of this animal (see Fig. 2 and Table S1 in Supplementary Information). The Euclidean distance between the model and sample value was then calculated for each measurement taken:

$$\sqrt{{(x-y)}^{2}}$$
(1)

In Eq. (1), x is the value of the measurement in the model and y is the value of the measurement in the sample.

The results obtained were normalized to fit the distances to a common scale, with all distances contributing equally to this assessment. Finally, the average of the normalized distances was calculated, providing an overall size and a similarity index. A high percentage indicates a similarity closer to the model. To select the most similar samples, a cut-off value of 75% was employed.

Secondly, the relative sizes of the figures produced in Tasks 1, 2 and 3 was evaluated to assess how experts and non-experts utilized the available space in the media. A statistical study based on comparison tests was used. Before running the comparison tests, normality and homoscedasticity were assessed using Shapiro–Wilk test and Fisher’s test, respectively. For Task 1, as data was not normally distributed, the non-parametric Wilcoxon-Mann–Whitney test was used; while for Tasks 2 and 3, as data met the normality and homoscedasticity assumptions, Student's t-tests were used.

Finally, the presence/absence of relevant anatomical details was examined in the realistic image generation task (Task 1) and in the drawing and engraving from memory tasks (Tasks 2 and 3). The goal of these final formal analyses was to objectively assess the quality and accuracy of the generated images compared to the original model, examining differences between experts and non-experts in visual arts. For this analysis of the presence/absence of anatomical details, a total of 33 attributes and 79 values were defined (see Table S2 in Supplementary Information), which were subsequently analyzed using a Correspondence Factor Analysis (CFA)79 and a Cluster Analysis (Ward's method and Euclidean distance).

Regarding the technical analyses, in Task 3, we calculated a “technical analysis index” of the productions made with prehistoric tools. Specifically, the presence/absence of errors in the execution of the engraving from memory task was analyzed, assessing the ability to engrave without executing involuntary accidents, which could result from a lack of knowledge of the interaction between the burin and the media, as well as imbalances of the forces exerted45.

The motifs were observed using a Leica MZ 16 with 16:1 apochromatic zoom, 7.1× to 115× magnification, using an added-on Leica IC90E, and a Dino-Lite AD-7013MZT version 1.3.2 digital microscope, with 5-megapixel resolution and 10× to 250× magnification. Microscopic analysis of the lines was based on previous work44, which determined a series of technical indices or stigmas (i.e., microscopic indices that reveal the engravers' gesture)47 indicative of greater or lesser motor control of the tool and the engraver’s actions. There were six stigmas associated with expertise (i.e., depth of the incision; use of combined profiles using different active parts of the tool to generate different types of strokes; precision in the actions; differential relief; combination of techniques; and preparation of the surface) and six associated with non-expertise (i.e., difficulty in deepening a single groove; the tool going outside the line; inflexions in curved lines; inflexions in straight lines; slips; and rectifications) (see Table S3 in Supplementary Information).

With these indices of expertise and non-expertise obtained from the technical analyses, a database was generated using Livecode© software to describe the motifs engraved by the participants. Expertise indices were quantified positively, while non-expertise indices were quantified negatively. Consequently, the “technical quality index” for each motif was between -10 and + 10, depending on the degree of expertise of each engraver. This methodology follows a framework previously applied to archaeological record and experimental programs44.

Results and discussion

Biomechanics of gesture

The Principal Component Analysis (PCA) on the finger positions of the dominant hand of each participant highlighted the most active parts of the hand during the tasks. In Tasks 1 and 2, movement patterns mainly involved the flexion/extension of the proximal and metacarpophalangeal phalanges of the five fingers, with the middle finger being particularly active. There were few flexion/extension and deviation movements of the wrist, and abduction/adduction of each finger and the thumb were not significantly involved in the movement of the hand. Task 3 showed a different pattern, with noticeable change in the most frequently used anatomical angles (see Fig. 3). The index, ring, and little fingers dominated in flexion and extension of the proximal and metacarpophalangeal phalanges, as well as in the involvement of the wrist with flexion, extension, and deviation movements. As in the previous tasks, abduction/adduction of each finger did not seem to be involved in the movement of the hand. Additionally, there was limited involvement of flexion and extension of the carpometacarpal joint of the thumb in this engraving task. No significant differences between experts and non-experts regarding the biomechanic data in any of the tasks were found.

Figure 3
figure 3

Principal component analysis of the median times of each of the anatomical angles analyzed in the flexion and abduction movements, and heat map of the most used phalanges and joints during the execution of the tasks.

Full size image

Formal analyses of realistic image generation and drawing/engraving of Paleolithic motifs

Based on their size, the similarity index between the drawn figure and the model (a real horse and the horse depicted in the cave of Les Trois-Frères) showed differences between experts and non-experts in Tasks 1 and 2. Specifically, experts exhibited similarity percentages above 75%, while non-experts did not exceed 50%. However, in Task 3, the differences between experts and non-experts diminished, with neither group exceeding 70% similarity (see Fig. 4a). In other words, the results suggest that while artistic expertise was consistently associated with accuracy in size during drawing tasks, this relationship was not observed in the engraving task.

Figure 4
figure 4

(a) Similarity index between the model and the figures represented in Tasks 1, 2, and 3. (b) Size of the figures relative to the media.

Full size image

In analyzing the use of media in Tasks 1, 2 and 3, experts used a significant portion of the available surface area in Task 1, W = 229, p < 0.001, Task 2, t(31) = 5.44, p < 0.001, and Task 3, t(31) = 4.63, p < 0.001. As shown in the boxplots (see Fig. 4b), experts used a substantial portion of the available surface area for the drawing from memory tasks, occasionally leading to incomplete figures, whereas non-experts used only a fraction of the available media to frame the motif depicted.

The presence/absence of relevant anatomical details in the drawn/engraved figures differed between experts and non-experts in Tasks 2 and 3, but not in Task 1, a task primarily measuring visual long-term memory. Specifically, the CFA discriminated the details represented in the figures, such as the presence of cervical-dorsal curvature, anatomical details, and treatment of perspective, assessing the quality of the generated image compared to the original model. On the one hand, experts were associated with the correct perspective of the front and hind legs, as well as with the extension of the line of the abdomen in both the front and hind legs. On the other hand, non-experts were associated with straight biangular and oblique biangular perspectives of the front and hind legs (see Figure S4 in Supplementary Information). Therefore, our results demonstrate significant differences in the accuracy of the reproduction of the generated image of Paleolithic motifs based on artistic expertise (see Fig. 5).

Figure 5
figure 5

Examples of the horse representations made in Tasks 1, 2 and 3 by expert and non-expert participants. The difference between Task 1 (drawing of a horse from memory) and Task 2 (copy of the Les Trois-Frères model) can be observed, as well as the differences between experts and non-experts in the control of certain aspects of the representation such as the perspective in all the tasks.

Full size image

Technical analysis of the engraving of Paleolithic motifs

The "technical quality index" was calculated according to the presence/absence of errors in the engraving task (Task 3). As noted above, this index assessed the ability to engrave without executing involuntary accidents resulting from a lack of knowledge of the interaction between the burin and the media, as well as imbalances in the forces exerted 45. In our sample, there were no significant differences between experts and non-experts, with values between -5 and -8 (values could range from −10 to + 10). Common errors included tool slipping and sliding, difficulty in deepening a single groove, and snagging of curved strokes (see Fig. 6).

Figure 6
figure 6

Examples of engravings made in Task 3 by expert and non-expert participants, showing common errors such as the difficulty in deepening a single groove (ID2, 24, 25, 32), snagging in curved strokes (ID4, 24, 32), and lack of incisions depth (ID15, 7, 14).

Full size image

These findings revealed the high level of inexperience of our sample in Task 3, an engraving task specific to Paleolithic art. This is consistent with the analyses of the biomechanics of gesture that indicated fewer or no differences between experts and non-experts. Both the technical quality index and the gestures suggest that the specific learning required for Paleolithic engraving techniques may result in differences in engraving between experts and non-experts.

General discussion

This research explores the cognitive abilities and motor skills involved in the creation of Paleolithic art motifs and engravings by comparing visual art experts and non-experts. In Study 1, we used psychometric tests to analyze spatial ability, assessing spatial visualization, mental rotation and perceptual speed, as well as memory capacity, assessing short-term memory and working memory using verbal and visuospatial information. In Study 2, drawing/engraving tasks involving Paleolithic techniques were used that assess participants' visual analysis and skilled motor execution. The results revealed differences between experts and non-experts across several tasks, highlighting the abilities associated with artistic expertise in visual art.

Firstly, our findings from Study 1, derived through psychometric approaches, showed significant differences in spatial abilities between experts and non-experts in visual arts. Specifically, experts outperformed non-experts in spatial visualization and mental rotation abilities. These findings align with the limited previous research suggesting that artists possess enhanced capacities to manipulate and process visual information efficientlye.g.,23,31,35, as well as research concluding that the ability to generate and transform mental images is important for the visual arts18.

However, not all data consistently point to differences in cognitive abilities between experts and non-experts. Specifically, no significant differences were observed in perceptual speed or memory, neither for working memory nor short-term memory, regardless of whether the memory involved verbal or spatial information. Although previous research suggested that perceptual speed is a component of spatial ability52,53,54,55, in Study 1, perceptual speed did not differentiate between visual arts experts and non-experts, unlike spatial visualization and mental rotation. However, these findings align with more recent proposals by other authors like for example, Schneider and McGrew80, who argued that it is important to distinguish between speed-related abilities, like “speed of perception”, and level-based abilities, like “visual-spatial abilities”. In this case, a simple test such as the perceptual speed task can be easily solved by most participants, making it impossible to observe differences between our expert and non-expert groups.

Furthermore, previous literature suggests a relationship between working memory, spatial abilities, and processing speede.g.,81. Therefore, it was essential for us to confirm that there were no significant differences in memory performance between our sample of experts and non-experts in visual arts, as such differences could confound our understanding of the results related to spatial ability. Since there were no differences between experts and non-experts in visual arts in working memory and short-term memory, both for verbal and spatial information, we can confidently conclude that there are cognitive processes associated with the creation of visual art, emphasizing the importance of spatial cognition as the basis of expert knowledge in artistic expertise in visual arts.

Secondly, in Study 2, a new methodology was presented to evaluate the visual accuracy of copied images based on quantitative dichotomous and criteria. This methodology allows us to objectify the notion of accuracy in copying, as understood in previous works20,82,83. Likewise, using the Rokoko Smartgloves to measure dominant hand movement enabled the objective measurement of motor coordination.

The results showed that experts in visual arts accurately represented anatomical details and perspective in their drawings, unlike the more rudimentary productions of the non-experts. In other words, experts were closer to the real model in terms of the similarity index. This was evident in the analyses of the figures from the experimental tasks: drawing a realistic horse from memory (Task 1) and drawing and engraving a horse figure depicted in Paleolithic art from memory (Tasks 2 and 3, respectively). Additionally, experts utilized more surface area in all three tasks, reflecting their strategic approach to artistic expression and superior control over the media. These findings highlight the precision and strategic space use required for artistic expertise in visual art. As Gustavsen84 noted, novice artists often do not consider the distribution of space, focusing primarily on the most representative aspects of the figure. Finally, we observed that experts displayed greater visual memory, accurately reproducing aspects of the original model, such as perspective or animation, while non-experts showed difficulties with these elements. This aligns with previous research indicating that artists have a higher capacity for local processing of visual details22,82,85 and better encoding and depiction of key object features20,25,35,37,86,87. Chamberlain et al.88 also found a significant correlation between visual memory and drawing ability, implying the involvement of visual memory in the drawing process.

Despite the differences found between experts and non-experts in the figures depicted in the experimental tasks, no differences were found between the groups regarding the biomechanics of gesture in any of the tasks. This suggests similar motor skills between experts and non-experts in visual arts, as has been also revealed by previous studies89. These findings support Cohen and Bennet's90 conclusion that differences in visual accuracy between expert and non-expert artists are due to non-artist's misperception of the object, rather than differences in motor skills. In our Study 2, the absence of differences in motor skills in the drawing tasks (Tasks 1 and 2) could be explained by the familiarity with the drawing technique, which lacks specific biomechanical characteristics and is commonly learned in Western societies from childhood90. Regarding the engraving of a figure depicted in Paleolithic art (Task 3), the lack of differences in motor skills between groups may be due to neither group being familiar with this specific Upper Paleolithic technique. This aligns with previous studies45, indicating that specific training in Paleolithic engraving technique is needed, regardless of drawing talent or formal art education. This is supported by the low technical index evidenced by the works on slabs from both groups. This low technical index is also present in the archaeological material, specifically in those works exhibiting the same technical stigmas that reveal difficulties in handling the tool. The convergence of the experimental protocols with the archaeological evidence reinforces the hypothesis of an artistic apprenticeship in the Paleolithic44,47.

Taken together, this research found that artistic expertise in visual art is associated with better performance in some spatial abilities, with experts showing higher mental rotation and spatial visualization than non-experts (Study 1). Additionally, artistic expertise is associated with a specific pattern in drawing from memory, with greater accuracy in the size and formal details of the drawn figures in experts than non-experts and better utilization of the media size in all tasks (Study 2). These results support the idea of a consistent relationship between drawing from memory and cognitive abilities in the context of artistic expertise. Kozbelt20 conducted one of the most extensive analyses of the differences between artists and non-artists, finding that artists scored higher than non-artists in both mental rotation and drawing tasks, among others. Additionally, these tasks were highly correlated. Goldsmith et al.91, later confirmed these findings using drawing, mental rotation, and spatial visualization tests, showing that participants with low spatial ability also scored significantly lower in drawing tasks. They argued that participants with high spatial ability could represent what they saw more accurately. Thus, previous literature, in line with our findings, highlighted that artists exhibit unique abilities shaped by their experience and expertise in visual cognition.

In summary, the results of the two studies provide compelling evidence that motor skills and spatial cognitive abilities are intrinsically associated with artistic expertise. By analyzing the cognitive and motor processes underpinning artistic expertise, our research contributes to a deeper understanding of the processes involved in Paleolithic artistic production. However, this study also highlighted that Paleolithic art requires a specific technical apprenticeship related to motor skills, corroborating previous findings44,45. Thus, this research lays the groundwork for future research to determine which specific motor and cognitive abilities are linked to Paleolithic art and how they differ from those attributed to artists in general, beyond the role of innate talent39.