Affordances and curvature preference: The case of real objects and spaces

Chuquichambi and colleagues recently questioned the prevailing belief that a universal human visual preference exists for curved shapes and lines. Their comprehensive meta‐analysis demonstrated that while curvature preference is widespread, it is not universally constant or invariant. By revisiting their dataset, we made an intriguing discovery: a negative relationship between curvature preference and an object's “affordances.” Taking an embodiment perspective into account, we propose an explanation for this phenomenon, suggesting that the diminished curvature preference in objects with abundant affordances can be understood through the lens of embodied cognition.

the visual domain is general and common, though not universal and invariant." 1 Chuquichambi and colleagues provide a valuable contribution by questioning the universality of the preference for visual curvature.
In this commentary, we develop a specific argument, grounded in their database of prior study results, that the curvature preference can also be linked to the "affordances" of the presented objects.The concept of affordances refers to the possibility of interactions with an object. 2 Affordances have been defined as, "the possibilities for use, intervention, and action which the physical world offers a given agent . . .determined by the 'fit' between the agent's physical structure, capacities, and skills and the action-related properties of the environment itself." 3 As we will show in the following discussion, preferences for visual curvature tend to decrease as the affordances provided in the research studies increase.To demonstrate this effect, we focus on prior studies from Chuquichambi and colleagues' database that made use of virtual reality (VR) or real objects-as compared to those that merely presented participants with pictures of objects.The immersion created through either VR or the use of real objects induces an (A) (B) (C) Comparison of Hedge's g in studies of VR or real-world objects with different levels of affordance.(A) Comparing the effect size of the studies that made use of images as stimuli with studies that made use of VR or real objects.The median of the effect decreased from images, being 0.26, to VR or real objects, being 0.19.(B) Within the group of VR or real objects, three groups were further identified based on their affordances.Once again, as the affordance landscape becomes richer, the effect of preferences for curved objects begins to vanish.(C) A similar pattern is found in the studies that investigated food preferences.Data in all charts are from Chuquichambi and colleagues. 1 The dotted yellow line indicates the median.Due to great differences in the number of studies in each group, the median has been used to reduce biases.Abbreviations: n.s., not significant; VR, virtual reality.*p < 0.05; ***p < 0.001.
intrinsically greater perception of affordances (possibilities for interaction) compared to looking at two-dimensional pictures. 4 further provide in this commentary an embodiment perspective for why the curvature hypothesis does not extend to objects rich in affordances.Based on an enactive and ecological outlook, we briefly discuss the nature of the stimulus in geometry preference experiments, which we believe to be of crucial importance in the study of neuroaesthetics.Our findings demonstrate the importance of ecological validity in experimental setups, showing that researchers should be cautious about generalizing results based on artificial laboratory settings or image representations to real-world contexts.

SHAPE PREFERENCES AND OBJECT AFFORDANCES: A COMPARISON OF EFFECT SIZE
Using the same database as Chuquichambi and colleagues, 1 we filtered the studies to exclude those that used photographs or images.
This left a total of 12 studies that presented participants with VR or real objects.Based on the prior literature, 4 we assumed that these 12 studies all provided greater affordances than those based on images.A comparison between studies that made use of images and VR or real objects revealed a slight decrease in the median, which is also reported by Chuquichambi and colleagues (Figure 1A).An unpaired t-test with unequal variance revealed a slightly significant difference between the two conditions.However, due to the limited number of studies, the statistical comparison should be taken lightly.The main point of the comparison is to visualize the differences in effect size once richer affordances are introduced.We thus also analyzed the extent of affordances provided by each VR or real-object study, in relation to the effect sizes that the studies found for curved-object preference.The affordances were divided into three categories, including those that presented objects that invited manipulation by hand, those that invited touching and tasting (food items), and those that invited physical movement and exploration (immersive indoor environments).At the first level of affordance, consisting of hand manipulation only, preferences for curved shapes remained definitive, with solid effect sizes.For the studies with higher levels of affordance, however, this preference for curved shapes vanished (Table 1 and Figure 1B).An unpaired t-test revealed significant statistical differences between these conditions.
In regard to specific studies, we categorized Alp's 5 findings as "hands-only" since this research focused on a scale model rather than an actual immersive interior.In this study, a group of chemists and architects were given miniature models (1:20) of curved and angular spaces and were encouraged to manipulate and adjust the models as they pleased to create a better understanding of the interiors.Physical objects at this size essentially appeal to the motor skills of hands, rather than to walking or other full-body movements.The affordances were similar in the studies by Jakesch and Carbon, 6 Soranzo and colleagues, 7 and Gao and Soranzo, 8 which made use of physical objects of a size that would fit into the hands.It is worth noting that the studies by Alp, by Soranzo and colleagues, and by Gao and Soranzo, all placed an overt emphasis on affordances by encouraging the participants to manipulate the object or to consider doing so.Soranzo and colleagues, 7 for example, instructed participants to consider the question "what do the objects do?"Despite this emphasis on engaging with and using the objects (rather than just visual appreciation), all of these studies found evidence of preferences for curved forms with solid effect sizes.This indicates that the affordance level of hand-based manipulation is not sufficient to reduce the observed visual preferences for curved forms.
The situation changed profoundly when we examined studies with affordances that appealed to the gustatory system.When participants were presented with real-world objects associated with eating, there was no longer any distinguishable preference for curved geometries (Figure 1C).An unpaired t-test revealed slight but significant differences between the conditions.The studies in this category evaluated curved versus angular shapes in the context of plates, cups, food labels, or food items.We regard such objects as presenting a higher level of interaction affordances since they invite participants to consider bringing the objects to their mouths and using them in the process of ingestion.
Finally, our third category of affordances included studies that invited participants to consider full-body movements and the exploration of an environment.Unlike those in the previous categories, these studies made use of VR environments rather than real objects.
However, we regard the immersive VR contexts as having more similar affordances to real-life stimulus rather than to picture-based studies, since in VR research the participants can engage in direct interactions and manipulation of the presented objects/environment.
The relations between virtual and real environments have been assessed elsewhere, 9,10 with minimal differences found in physiological measures and spatial reactions.In this context, Tawil and colleagues, 11 to their surprise, were unable to identify a curvaturepreference effect when participants were allowed to freely roam around a VR interior environment with angular versus curved features.Shemesh and colleagues 12 had similarly inconclusive results.
Such interior environments are regarded as quite rich in affordances because they are designed to support human activities and include many opportunities for interacting with objects and features. 13, 14 In summary, although Chuquichambi and colleagues 1 convincingly found an overall pattern of human preferences for curved geometries in the prior literature, our analysis shows that this effect dissipated when participants were invited to consider closely interacting with the presented objects, rather than just viewing them in a detached fashion.We believe that this issue of affordances in relation to geometry preferences raises important questions about the ecological validity or generalizability of the prior research literature.

STIMULUS AND ECOLOGICAL VALIDITY: AN EMBODIED PERSPECTIVE
To systematically investigate how the design of objects and of the built environment affects human experience, it is highly important that the stimuli being evaluated are presented in an ecologically valid manner.
The concept of stimulus is central to the study of neuroaesthetics; yet, the question of what is included in a stimulus is frequently overlooked.
Visual representations that are recognized by participants as being pictures may not carry the same impacts as actual objects that have the potential for interaction. 21,22ersimplifying the nature of the stimulus is not a novel issue in without an environment surrounding it.Equally, although not so obvious, an environment implies an animal (or at least an organism) to be surrounded." 2 The extent to which an experiment's findings may be generalized to real-world situations, or the degree to which a study represents the complexity and variety of the actual world, is referred to as "ecological validity."Affordances are important in ecological validity because they relate to the potential activities that an object or environment permits an organism to do.The idea of affordances is important in designing ecologically viable studies because affordance perception is a key component of everyday behavior, and the capacity to detect affordances in a laboratory context is critical for duplicating realworld behavior.Maintaining an approximation of ecological validity in research settings is no simple task, but it is an endeavor worth foregrounding.According to the embodied perspective suggested here, action is not an additional parameter but an inherent part of perceiving and experiencing the world, including our reactions to the shapes and surfaces within it.When an organism perceives affordances in the environment, the organism perceives the environment and the self as a whole, entirely embedded within its system. 24Thus, ecological validity is tightly linked to understanding the physical structure of the organism's body and all of its relevant sensory, cognitive, and activity potentials.Such an outlook on human embodiment has also sometimes been called "enactivism," holding that interactions with the world are decisive for human experience, cognitive performance, and behavioral decisions. 25,26[29][30][31] If the curvature-preference effect were to be consistently identified in research settings as rich and complex as the actual built environment, then it would suggest that such responses may indeed be universal.However, the studies evaluated by Chuquichambi and colleagues 1 show a strong reliance on viewing photographs or images in a detached fashion; and in the smaller number of studies that did have more robust affordances, this curvature-preference effect could not be found.The embodied-based perspective argues that more realistic immersion and the associated affordances (ability to interact) are critical to the human experience of geometries in actual architecture. 27,31,32It is, therefore, an intriguing possibility to consider that strong curvature preferences might be limited not only to visual preferences but specifically to preferences when viewing pictures.
Human reactions when viewing actual three-dimensional architecture in real-life, or even in VR contexts, may well be different since the affordances for action in those contexts are extremely different.Interacting with photographs is not ecologically comparable to interacting with the built environment.Indeed, the results of the analysis presented here seem to indicate that preferences for curved shapes do not hold up in high-affordance environmental contexts.
It is sometimes assumed that an effect seen in the lab, under mild or controlled conditions, will only be more pronounced when it is encountered in a "raw" or real-world form, where organisms generally have more at stake.The data, in this case, suggest that the opposite can also be true and that effects seen in the lab can dissipate in more immersive environments.The dissipating effects seen in the current meta-analysis may likely be related to the limited affordances of the lab environment and to the richer and more complex experiences associated with multisensory stimuli when experiments move beyond the use of 2D pictures.
When it comes to preferences for curved geometries, we believe that the data speak more to a phenomenon associated with detached or secure viewing conditions with a fixed point-of-view and severely limited interaction possibilities, rather than being a universal outcome of design.Subsequent work by Munar, Chuquichambi, and colleagues 33 was able to identify a preference for curvature when looking at paintings in real-world museums, but this again references 2D pictures in a detached viewing environment, which retains many aspects of the laboratory context in regard to the affordances that are available when viewing the painted objects.As Chuquichambi and colleagues 1 briefly and rightfully touch upon near the end of their meta-analysis, an empirical neurobiological investigation of the visual curvature hypothesis versus contextual affordances might help us to more precisely characterize and understand the nature of this preference.Such studies might also expand upon the range of metrics that have been used in this research area for evaluating "preference" responses, such as studying brain reactions via electroencephalography or measuring behavioral reactions, rather than relying solely on self-report instruments.

CONCLUSION
Among the first researchers to claim a biological raison d'etre for curvature preferences were Bar and Neta. 34,35Elaborating on their findings from functional magnetic resonance imaging studies, they proposed that humans responded more positively to curved geometries for evolutionary reasons-arguing that angular contours were associated with dangerous objects, while curved forms were more frequently associated with safety and comfort.Like most such experiments, Bar and   Neta's participants were studied while viewing pictures of objects, in this case, while laying down in a noisy scanner.From the embodiment perspective, the affordances of this experimental setup were profoundly incompatible with the "natural circumstances" in which the pictured objects would ordinarily be encountered.Although the current Commentary should be understood as an exploratory post-hoc analysis and interpretation, we are disinclined to agree with a (speculative) evolutionary psychology interpretation of such data. 36,37It seems fairer to suppose that, just as a knife may be considered either a valuable tool or a dangerous threat in different contexts, humans in natural environments may have widely differing reactions to various geometries depending upon their specific embodied situation.
In agreement with Chuquichambi and colleagues, 1 we believe that numerous confounding variables may be related to participants' expressed geometry preferences, including aspects of the experiment's design as well as a variety of participant-specific factors.Without a better accounting for such variables, it would be remiss to draw universalizing conclusions.From our current analysis, it appears that the extent of affordances in prior experiments may play a central role in the presence or absence of a curvature preference, but this is likely just one variable among many.We would like to suggest that the primary insight to be gained from the meta-review and dataset is that there is a pressing need for researchers to move as much as possible toward more ecologically valid experimental setups.As has been suggested elsewhere, [38][39][40] there is also a need to integrate ecological validity with robust behavioral and brain-imaging techniques to improve our scientific understanding of human responses under natural conditions.It is becoming clear: for studying cognition and behavior, the affordances and context matter.

AUTHOR CONTRIBUTIONS
Z.D. conceived the idea.Z.D. performed the analysis.Z.D. and S.K. interpreted the results.Z.D. and S.K. drafted the manuscript.Z.D. and S.K. edited, revised, and approved the final version of the manuscript.
science.It is a recurrent one.Many years ago, it was directly tackled by Brunswik,23 who cautioned against generalizing from laboratory studies and developed a "theory of the environment" that emphasized ecological validity in experimental designs.According to Brunswik, the organism and phenomenon in question must be studied under their "natural circumstances," which too often have been reduced for research logistics.While the findings from laboratory studies may be quite robust, the problem arises when they are generalized to realworld contexts where the stimuli may seem superficially similar but are actually quite different.Gibson 2 developed the concept of affordances based in part on this same concern, arguing that perceptions of the environment become meaningful and induce certain responses only in the context of the range of potential actions that the participant can undertake.One important strength of this ecological insight was in overcoming false dichotomies, such as action versus perception, organism versus environment, and subject versus object.Instead, a relational and systems view emerged in which action and perception are considered to be part of a single, interconnected reality.As Gibson stated, "Each term implies the other.No animal could exist Affordances and shape-preference conclusions in 12 studies that made use of VR or real objects rather than 2D pictures.
TA B L E 1Note: The shades of gray indicate increasing levels of affordances in the studies.Only one-third of these studies revealed a preference for curved shapes.Abbreviation: VR, virtual reality.