Abstract
Perceptual insight, like recognizing hidden figures, increases the appreciation of visually perceived objects. We examined this Aesthetic Aha paradigm in the haptic domain. Participants were thinking aloud during haptic exploration of 11 visually nonaccessible panels. They explored them again evaluating them on liking, pleasingness, complexity, and interestingness. Afterwards they rated photographs of the panels on the same variables. Haptic pleasingness was predictable by the strength of insight (Aha!) during free exploration and the material feel. Liking was increased when complexity was high in addition. Pleasingness and interest were negatively related to each other but predicted liking in a combined model. Personality and explorative strategies were considered, for example, strength of insight was increased for ambiguity-tolerant people, and people with high need for closure explored more globally. Evaluations of haptic and visual explorations correlated significantly, and in both modalities, complexity correlated more strongly with interest than with liking. Our study transfers the Aesthetic Aha effect to the haptic domain and reveals slight differences in its hedonic quality with a potentially higher relevance of pleasingness. We suggest that revealing a (meaningful) structure during exploration—visually or haptically—can enhance positive affect and interest hereby benefits from an increased level of complexity.
Introduction
We are motivated to explore something when we appraise it to be challenging and at the same time see the potential for new experience (Muth & Carbon, 2016a) or coping (e.g., Berlyne, 1971; Silvia, 2005). And recognizing a hidden figure in a visual pattern increases the appreciation of the image reflecting our tendency of and joy in revealing perceptual structure (Muth & Carbon, 2013). Can these effects be transferred to the haptic domain? The systematic pattern of abrasion by repeated touches of a relief as shown in Figure 1 is a sign of curiosity and attraction. Despite the potential regress biasing people toward spots due to traces of older touches, the specific pattern of abrasion does not seem random at all but linked to semantics. This might underline the importance of touch for understanding and for gaining access to an objects’ presence. The haptic sense might be largely underestimated with regard to its role in providing us with an (intellectual) grasp of the world. The present article therefore examines affective responses to insight during haptic exploration.
Detail of the gates of the Milan cathedral (planned by L. Pogliaghi) with abrasions resulting from touching behavior (see hands). Photograph by Claus-Christian Carbon. This figure appeared first in Carbon (2011) under common rights according to CC BY-NC-ND 3.0.
Characteristics of the Visual and the Haptic Modality
Touching the relief involves mechanoreceptors and thermoreceptors in the skin and kinesthetic or proprioceptive receptors in muscles and tendon organs (Lederman & Klatzky, 2009). Active touch therefore comprises several stimulus qualities: hardness of material, temperature, surface, shape, size, weight, and balance (Sonneveld & Schifferstein, 2008). An event mostly affects various sensual modalities more or less simultaneously (besides mode-specific temporal differences), and reciprocal intermodal influences are evident from many illusory phenomena (e.g., visuo-tactile-proprioceptive interaction inducing the rubber hand illusion; Botvinick & Cohen, 1998; or the audiovisual McGurk effect; McGurk & Macdonald, 1976). After all, “we do not see one world of sight, hear another of sounds, taste or smell another … Instead, we know one world, a world of sunsets and trees … ” (Marks, 1978, p. 185). Sensory substitution experiments nicely reveal how perceptual habits are formed by interaction: Blind or blindfolded participants feel the response of a device that translates optic stimulation into tactile stimulation on their skin at first. If they are given the opportunity to manipulate the object, they sometimes perceive a distal object in space instead after a while—like when we use a cane and experience interaction not at the palm holding it but on its end (see for a discusion of these studies Gapenne, 2013, hereby pointing to the role of proprioception). Rare situations, like when we have an aesthetic experience, allow (re)gaining access to the quality of the sensual modality which is often almost transparent due to this objectivation (see also ideas by art theorist: Fiedler, 1971). We cannot, for instance, transfer the experienced sound of a piece of music into other modalities. An enactive perspective focuses on the modality-specific pattern of interaction, the so-called sensorimotor contingencies. When we move, auditory stimulation changes in a different way than tactile or gustatory perceptions and “as the eye moves, the resolution of the sampled information changes in a lawful way which is typical of the visual modality” due to the inhomogeneity of receptor distribution (O’Regan & Noë, 2001, p. 85).
Our ideas of the specifics of each sensual modality as well as their hierarchization are therefore partially rooted in their embodiment and the specific patterns of interaction. According to Jütte (2000), they are also culturally constructed and influenced by technological inventions like the modern letterpress by Johannes Gutenberg in 1450. Whereas already in the Greek and Latin era, vision was described as the first sense by many scholars in the ancient world, the haptic sense alternates between extremes: On one hand, it was described as first in the row given its distribution over all body parts, its role as a basis for all other senses (an idea that was also discussed later-on, see Marks, 1978), its link to intellectual sensitivity and its relevance for survival. On the other hand, especially since the 13th century, it was mentioned last in the row due to an association with subordinate feeling (as opposed to reason), shame and sin mediated through its link with erotic touch (according to Jütte potentially also linked to Eve’s touch of the apple before seducing Adam). Vision traditionally had a prominent role given the link to the typically human upright position enabling a wide view, links to language often expressing understanding and reason (like I see or to foresee; note that this could be a mutual causality), the ease in sharing what one sees with others (as compared with the proximal senses), the speed of perception as well as the possibility to see distant objects that are inaccessible by other senses. (Jütte, 2000).
Haptic perception is dynamic and requires motion—but this also accounts for the exploration of visual scenes, for example, due to the inhomogeneity of the receptor distribution on the retina (Findlay & Gilchrist, 2009; Noë, 2005). Still, the eyes’ position in relation to each other is quite rigid compared with the higher flexibility of the spatial relation between body parts that can be involved when we touch something (Overvliet, Krampe, & Wagemans, 2012; Overvliet & Sayim, 2016). Furthermore, the shape of objects and the relations between elements is caught faster by vision than by haptic perception due to differences in the size of the perceptual fields (Overvliet & Sayim, 2016). Haptic perception might be unique due to its reciprocal character “Seeing does not imply being seen, neither does hearing imply being heard. But touching implies being touched simultaneously” (Sonneveld & Schifferstein, 2008, p. 41). The perception of our body in space by touch and proprioception might even be the base of each interaction establishing perception (see Gapenne, 2013 for the crucial role of proprioception). Furthermore, we frequently use touch to estimate authenticity (fake material or real material; Carbon & Jakesch, 2013), and people vary with regard to their need for touch (Peck & Childers, 2003) to make decisions about objects and build preferences.
We are interested in similarities between vision and touch with regard to the Aesthetic Aha, a hedonic response to Gestalt detection. Gestalt psychology suggested principles guiding the perceptual grouping of elements like when we perceive a coherent pattern visually popping out of its background due to the similarity of its elements in color or shape or the salience of a melody independent of key or instruments (e.g., Wertheimer, 1923). These ideas from the beginning of the 20th century were extensively examined ever since. The question remains which stages in perception are affected by Gestalt principles, how they are dynamically connected (feedforward or feedback) and whether sensual modalities differ in this regard (Overvliet et al., 2012; Palmer, Brooks, & Nelson, 2003). An integration of global features also happens in the haptic domain as evident from modulations of context effects by grouping of distractors (Overvliet & Sayim, 2016). We also find some evidence for grouping by similarity of surface roughness, especially if the task is of high complexity (Van Aarsen & Overvliet, 2016). Good continuation applies as well but not ameliorated grouping by proximity (Overvliet et al., 2012)—the latter being meanwhile beneficial for contour detection (Overvliet, Krampe, & Wagemans, 2013). Given the evidence that similarity is of help for the grouping of haptic elements, it seems probable that Haptic Panel 10 in Figure 2 allows for the detection of the shape of fishes but also of the big emergent fish given the similarity of its constituting elements. And the carapace of the turtle in Panel 5 might appear segregated from its body given the similar texture characteristics of its parts contrasted with the texture of the remaining body. It is evident that suitable movements are necessary for such Gestalt principles to become effective. There are multiple explorative strategies that perceivers use depending on context and goal of the exploration. The seminal work by Lederman and Klatzky (1993) examined so called exploratory procedures (see stage of exploration in Carbon & Jakesch, 2013). These are patterns of hand movements which are spontaneously applied in order to detect objects’ features, for example, lateral motion to extract texture and contour following to extract global and exact shape. In general, people show quick recognition capability when haptically exploring familiar objects (mostly within 3 seconds, Klatzky, Lederman, & Metzger, 1985), especially in case of a highly salient feature (like the cold smooth feel of your key among other objects in your bag). Whether they choose a slow, serial search strategy or a coarse handswipe depends on the contrast between targets and distractors regarding haptic features (e.g., roughness and therefore friction) potentially evoking a pop-out effect as known from visual search (Plaisier, Bergmann Tiest, & Kappers, 2008). In such a case, detection is fast, whereas raised line drawings without much variation in other features (texture, temperature, etc.) are very challenging (see e.g., Wijntjes, van Lienen, Verstijnen, & Kappers, 2008).
Black and white photographs of the 11 haptic panels utilized in the present study. The original story describes how a little black fish meets marvelous sea creatures after his school of “red friends” had been eaten by a big fish. When he meets another group of frightened fishes, he suggests a creative idea: They form the Gestalt of a big fish together with himself as the eye to be safe when meeting big fishes (the story was not told to participants).
Affective Qualities of the Visual and the Haptic Sense
It was repeatedly hypothesized that a detection of meaningful structures could induce positive affective responses potentially motivating us to explore novel or uncertain situations (Biederman & Vessel, 2006; Muth & Carbon, 2016b; Ramachandran & Hirstein, 1999; Van de Cruys & Wagemans, 2011). Indeed, there is evidence for affective responses to the appearance of Gestalt as well as to the mere promise of Gestalt within visual patterns. Gestalt detection within perceptually challenging images triggers an increase in liking; we called this earlier the Aesthetic Aha effect (Muth & Carbon, 2013). Corresponding evidence can be found for visual patterns (Muth & Carbon, 2013; Stevanov, Marković, & Kitaoka, 2012) as well as for perceptual insights during an artistic movie (Muth, Raab, & Carbon, 2015, 2016). The effect appears to be independent of the valence of the motives (Chetverikov & Filippova, 2014). It might not be the ease of processing that is relevant for an increase in liking here (as reported for pleasure responses, e.g., Graf & Landwehr, 2017; Reber, Schwarz, & Winkielman, 2004) but rather the reward by gaining insight into a (meaningful) structure. We can similarly separate simplicity from order and complexity from disorder: Liking of a visual pattern was best (and positively) predictable by considering its orderedness, not its simplicity (Muth, Westphal-Fitch, & Carbon, 2018). People can enjoy complex order or unity in variety (see review by Van Geert & Wagemans, 2018) and also the Aesthetic Aha is pronounced in perceptually challenging contexts (Muth et al., 2016). The strength of cognitive and perceptual insight is even effective in cases of visually indeterminate stimulation that is not resolvable by clear Gestalt identification—as it is frequently the case when perceiving art (Muth, Hesslinger, & Carbon, 2015). It still needs to be clarified how close such insights have to resemble a Gestalt detection in the strictest sense and how far they can differ from determinate perceptual solutions and coherent interpretations.
Interest is strongly related to the motivation for exploration. An unfulfilled but promised insight could trigger an orientation response and a drive to explore (Muth, Raab, et al., 2015). This was already suggested by Berlyne (1971) linking the concept to collative variables like ambiguity and novelty and proposing that interest reflects complexity together with a promised perceptual success. More recently, interest was examined within the realm of appraisal theories of emotion and similarly linked to a combination of appraised novelty or complexity and coping potential (Silvia, 2005). It could indeed be shown that interest is increased before Gestalt detection happens (Muth, Raab, et al., 2015). Other findings show that interest is evoked by visual patterns of complex order (Muth et al., 2018) and people show interest for disturbing art (Turner & Silvia, 2006). Interest is characterized by longer viewing times (Berlyne, 1971; Silvia, 2005) and is decreased if ambiguity is easily dissolvable in art (Muth, Hesslinger, et al., 2015).
Empirical studies of hedonic responses to perceptual characteristics or processing characteristics are mostly based on vision. An exception is a well-known effect in the visual domain, the Mere exposure effect (Zajonc, 1968). It predicts an increase in liking with familiarization especially for complex visual stimuli, evoked by repeated unreinforced presentation. This effect was also shown for the haptic exploration of complex stimuli made of stone but not for others made of wood (Jakesch & Carbon, 2012). At the same time, haptic perception might involve specific affective dimensions. Disgust, fear, comfort, and pleasingness might for instance play a more dominant role for a proximal (and reciprocal, see earlier) sense like touch than for vision as exposing one’s body to direct contact can be harmful but also comforting, warming, sensual (e.g., erotic), and so on. The traditional association between touch and intimacy and its distinction from intellectual engagement therefore appear plausible although both require empirical examination (Jütte, 2000; on a sidenote: feeling touched might express such emotional intimacy).
It therefore seems fruitful to have a closer look at the hedonic dimension of pleasingness especially when it comes to touch. Clear conceptualizations and differentiations between pleasure and liking are rare (as are those between pleasure, pleasantness and pleasingness; please therefore excuse the corresponding inconsistency in the following section). But pleasure and interest were described as distinct routes to liking with pleasure being a hedonic response during automatic processing, and interest being raised when rather in a controlled mode of processing (e.g., when reducing disfluency in perception, Graf & Landwehr, 2015). Interest itself is not a deliberate process but related to an involuntary increase in motivation to explore. Smith and Ellsworth (1985) asked people to evaluate memories linked with various emotions via eight appraisal dimensions. Feeling interested was associated with more anticipated effort and less certainty about a situation than happiness. The wish to pay attention was a distinctive feature of experiencing interesting situations. Whereas the authors link interest with pleasantness, Silvia (2005) and Turner and Silvia (2006) show that unpleasant events can be interesting.
In case of touch, pleasingness could refer to a hedonic response to several qualities like texture (e.g., softness), shape (e.g., roundness), and thermal aspects (e.g., not too cold). There is evidence for a positive response to smoothness (Peck & Wiggins, 2006), whereas roughness of sandpaper increased the estimated difficulty of a social interaction described in a story that was unrelated to the haptic task (Ackerman, Nocera, & Bargh, 2010). Pleasant touch was found to influence consumer behavior (see Carbon & Jakesch, 2013), and product quality was estimated higher when being of pleasing haptic characteristics (Grohmann, Spangenberg, & Sprott, 2007).
We asked whether effects of insight and complexity on liking and interest are evident for haptic exploration as well despite its often proclaimed distance from processes of understanding. Given the specifics of touch regarding its (dis)pleasing qualities, we focused on the dimension of pleasingness in addition.
Rationale of the Present Study
In a study involving quantitative and qualitative approaches, we examined if the Aesthetic Aha effect as well as the link between complexity and interest that were previously reported for the visual domain are transferrable to the haptic domain. Given the differences in sensorimotor patterns, physiology and historical reception of the two modalities, it is plausible to assume differences in affective dimensions being involved. An oversimplified guess would link touch with more intimate dimensions like (un)pleasingness, whereas vision might be connected with affective dimensions including reward by intellectual engagement and insight with pronounced effects on liking and interest. Indications against such a sharp differentiation concern the accuracy of haptic object identification, the relevance of several Gestalt laws in the haptic domain as well as the observation that people have a need for touch to judge authenticity and preference. We can therefore anticipate a close relation with understanding and potential corresponding effects on liking and interest for the haptic domain as well.
In order to examine the relevance of strength of insight for liking, the role of pleasingness as well as the relation between complexity and interest, we asked people to explore, describe, and evaluate haptic panels. We furthermore included measures of personality and examinations of the strategy of exploration. We assumed more local exploration when pleasingness is judged due to the relevance of the material’s texture. The strength of an insight is not equivalent to the detection of Gestalt (e.g., when it is very easy or when a pattern is registered that was detected already earlier) but might nevertheless benefit from contour following (see also Lederman & Klatzky, 1993). We will also examine the relevance of vividness of a person’s visual impressions which was previously reported to influence recognition of raised line drawings in a positive way (Lebaz, Jouffrais, & Picard, 2012). Need for closure was described as a desire for, for example, order, predictability, and structure and an aversion against ambiguity (Webster & Kruglanski, 1994). It might therefore be related to a rather global exploration strategy and potentially to stronger insights as well (due to the urge to reach closure). Similar effects might be observable for low ambiguity tolerance, the capacity to perceive and appreciate contradictions, inconsistencies, and complexities—especially for seemingly unsolvable problems and for new experiences (see inventary by Reis, 1996). Given its relevance for decisions about objects and preferences, need for touch (Peck & Childers, 2003) is an interesting measure that we will examine in an explorative manner expecting it to be linked with intense exploration and pleasingness.
Methods
Participants
Twenty-one participants (seven men, Mage = 22.2 years, SDage = 2.7, range = 19–30 years) took part in the study; 17 of them for course credit in psychology, all others without any compensation. They were naïve to the purpose of the study, and all of them gave written consent to participation and to video-recordings of their voice and their hand movements.
Stimuli and Apparatus
Eleven haptic panels (297 mm × 420 mm, see Figure 2) were specifically created for this study, based on illustrations by Leo Lionni from the children’s book Swimmy. The haptic panels consisted of selected combinations of the following materials: rug (body of turtle), fabric (eye of turtle), rough sandpaper (body of big fish), moderately rough sandpaper (stones on ocean bed), less rough sandpaper (ocean bed), plastic (eye of big fish and of small fish Swimmy), felt (algae), aluminum sheet with structure (body of small fish Swimmy), plastic foil (body of jellyfishes), rubber laces (tentacles of jellyfishes and feet of crabs), cork (crabs’ eyes), laminated cork (crabs’ bodies), mussels (mussels on ocean bed), smooth foamed rubber (small fishes and carapace of turtle), and ripped foamed rubber (teeth of big fish)—all fixed by glue upon cardboard. Each of these materials that were assembled in the haptic panels was evaluated on how pleasant it feels via a 7-point scale by an independent group of 10 participants; referred to in the following as material-raters (three men, Mage = 21.5 years, SDage = 2.1, range: 19–26 years).
In the main study, the panels were positioned one after the other in a box and were visually inaccessible for participants who explored them with both hands through two cutouts of a curtain. This exploration was observable from the back of the box (see Figure 3). The room was dimly lit to reduce interference on haptic impressions by strong visual stimulation. We decided against darkening glasses to create a comfortable and safe atmosphere; complete darkness might have induced fear to touch and discomfort due to the entire lack of visual control. The only light source in the room consisted of LEDs that were installed at the back of the box. They allowed filming of the hand movements with a camera being positioned at the back of the box focusing the inside of the box only, not the face of the participant. In addition, we recorded participants’ verbal descriptions with this camera. An additional visual stimulus set consisted of frontally shot photographs of these panel that were printed on A4 paper and mounted on cupboard (printed image size: 21.25 cm × 29.75 cm; see Figure 2). To exclude effects of color, we did not use the original panels as visual stimuli but used grayscaled versions of them. We furthermore handed out German versions of the following questionnaires:
Haptic box and participant’s hands exploring a panel (video-still recorded from the back of the box).
Need For Touch scale (NFT; German version by Nuszbaum, Voss, Klauer, & Betsch, 2010 based on Peck & Childers, 2003) with two subscales describing either the joy in haptic feedback and intrinsic motivation to touch (autotelic NFT) or rather goal-oriented exploration to gain information (instrumental NFT); the need for closure questionnaire (Webster & Kruglanski, including additional items on decisiveness by Roets & Van Hiel, 2007); the vividness of visual imagery questionnaire (VVIQ_PA; Version 2.1; Gruter, Gruter, & Carbon, 2008), comprising vividness of face geometry, face expression, landscapes, and landscape objects; the inventory for measuring tolerance of ambiguity (Reis, 1996) comprising four subscales including ambiguity tolerance for seemingly unsolvable problems, for social conflicts, with regard to the image of the parents, for role stereotypes, and for new experiences.
Procedure
All participants voluntarily agreed upon recordings of their hand movements as well as audio-recording during the study. They were given the instruction to think aloud during the haptic exploration of the panel. After this free exploration, they were asked to explore the same panel again four times evaluating it verbally on liking, pleasingness, complexity, and interestingness (in this order) via a 7-point scale. There were no time constraints for any of these exploration phases. After the last evaluation, the experimenter exchanged the panel and participants started the free exploration of the new panel, and so on. As the panels described a story altogether, they were presented in the same order for each participant following the narrative of the story. This potentially induced order effects; for example, by adaptation: Exploring a panel with curved shapes, for instance, influences the subsequent perception of a straight line (see Vogels, Kappers, & Koenderink, 1996, for haptic aftereffects). We nevertheless decided to use nonstandardized stimuli that potentially tell a story (with features like texture quality and complexity being inhomogeneously distributed) and to keep the storyline instead of a randomized order due to three reasons: (a) People will most probably build assumptions on the relations between the single panels, irrespective of their order. Keeping the order stable and the storyline reasonable (in cases when Gestalt is detected) reduces potential confusion by inconsistencies within the storyline that would potentially overwrite effects of insight and complexity; (b) Detection of more complex Gestalts, as in Panel 10 (many fishes forming a big fish), is more likely when their elements (e.g., single fishes) have been explored before; and (c) Providing a more entertaining context (by following a story) potentially reduces fatigue during the quite difficult task of haptic exploration and creates a more ecologically valid situation due to the meaningful episodic context. After the haptic explorations, the corresponding visual stimuli were rated on the same variables in the same order one by one (without free exploration). The whole procedure took about 90 to 120 minutes.
Results
Data preparation
Three independent raters, referred to in the following as insight-raters, assessed the think-aloud protocols recorded during the first free exploration phase of each panel on intensity of insight via a 7-point scale. We averaged these values for each stimulus. The raters furthermore reported if the participant described the texture of the panel (yes or no). We counted all trials for which at least two of the raters stated yes as ones in which the person was aware of the texture. Another group of three independent exploration-raters analyzed the videos of participants’ hand movements and estimated the amount of local exploration (when a panel was examined with small motion, without broader movements to integrate the sensations) versus rather global explorations (contour following and more holistic motion) for each trial. The differentiation was discussed with all three raters in a personal conversation. Note that the sum of both exploration strategies did not necessarily amount to 100% as the exploration strategies are not mutually exclusive and two hands can explore the panels in different ways in parallel. Of 1,155 explorations (21 Participants × 11 Stimuli × 5 Dimensions [free exploration, liking, pleasingness, complexity, and interestingness]), we excluded 443 explorations (none of them occurring in the free exploration phase) in which a participant did not touch the panel at all or for such a short time that at least one of the three raters could not estimate the explorative strategy. In addition, we created a material feel value for each panel. This value was based on averaged evaluations of how pleasant each of the materials felt that were combined in the respective panel as assessed by the independent group of material-raters (see Stimuli and Apparatus section). All data files are available via the open science framework (see https://osf.io/x4wq3/?view_only = 71c59f989f3346b2bbf9575d551ffb40).
In the following, we will describe results with regard to the Aesthetic Aha effect. We analysed if liking and pleasingness are predictable by the strength of insight as evaluated by the insight-raters and we examined the link between complexity and interest. We hereby consider the average material feel as evaluated by the material-raters. We furthermore present differences between tasks with regard to explorative strategies as evaluated by the exploration-raters and benefits of these strategies. Results of evaluations during visual explorations are added by comparisons to the haptic modality and finally, we present personality-related effects.
Aesthetic Aha and interest by complexity for haptic explorations
Average liking evaluations of a stimulus were not predictable by neither strength of insight (p = .95, R2adj = –.111) nor material feel alone (p = .25, R2adj = .025), also not by their combination as evident from a multiple regression (see Table 1 for all regressions models). If combined with complexity evaluations, the model predicted liking significantly with an explained variance of 72%. Pleasingness could be predicted by material feel alone but combined with strength of insight the model explained more variance (54%; see also Figure 4). Complexity did neither add any explained variance nor was it a significant predictor of pleasingness. Interest could be predicted by complexity evaluations with 86% of explained variance. Strength of insight and material feel did not add much to the model and were no significant predictors.
Results of Multiple Regressions on Liking, Pleasingness, and Interest.
Strength of insight and material feel predict haptic pleasingness. (a) Pleasingness of panels with regard to the average evaluations of independent insight- and material-raters as separate dimensions and (b) pleasingness with regard to the product of the average values of strength of insight and material feel.
Liking was also positively predictable by complexity alone (β = .64, p = .03), even if to less extend: F(1, 9) = 6.30, p = .03, R2adj = .346. As evident from Figure 5, this small effect might not be linear and is mainly driven by Panel 4 which is of very low complexity (most left data point). Meanwhile, the positive relation between complexity and interest is stable (β = .79, p = .006) even if we repeat the analysis without Panel 4: F(1, 8) = 13.46, p = .006, R2adj = .581 (but still note the inhomogeneity of the stimuli regarding their complexity). Average word count of the think-aloud protocols during free exploration was positively related to interest (r = .85, p < .001). A multiple regression analysis suggests that this relation is mediated by complexity as predictability of word count is nonsignificant (β = .28, p = .19) when considering complexity (β = .70, p = .008): F(2, 8) = 36.70, p = .001, R2adj = .877. Liking evaluations could be predicted positively by those of pleasingness and interest, whereas interest and pleasingness were negative predictors of each other when combined with liking (see Figure 6).
Interest, liking, and pleasingness for panels of varying complexity. Error bars represent ±1 SEM.
Results of three multiple regression analyses (every variable is handled once as dependent variable) combined in one diagram.
Explorative strategies
In order to reveal characteristics of participants’ explorative behavior during the different tasks, we compared the usage of local versus global exploration strategies based on estimates of three independent raters (see Data preparation section). A repeated measures analysis of variance on averages of these estimates with explorative strategy (global vs. local) and task (free exploration vs. complexity vs. interest vs. liking vs. pleasingness) as factors revealed significant main effects of explorative strategy—F(1, 10) = 41.82, p < .001, ηp2 = .807, with higher amounts of local (M = 41.30, SE = 3.73) than global (M = 26.73, SE = 1.86) exploration and a main effect of task—F(1.50, 14.99) = 47.34, p < .001, ηp2 = .826 (Greenhouse-Geisser corrected due to violations of the sphericity assumption) with more intense usage of both explorative strategies during free exploration than when posing a task (see Table 2 and Figure 7).
Average Estimated Amounts of Explorations During Each Task and for Each Explorative Strategy in %.
SE = standard error.
Estimated usage of explorative strategies (global vs. local) based upon analyses of video documentation of hand movements, significant results (p<.05, Bonferroni-corrected) are marked by an asterisk (for further simple main effects for each task see Table 3). Error bars represent ±1 SEM.
Furthermore, a significant interaction indicates that besides higher values for free exploration than for all other tasks, participants explored less globally during evaluations of pleasingness than during evaluations of complexity or interestingness. Furthermore, whereas local exploration strategies were dominant for each task, the difference to global strategy usage is biggest in case of pleasingness evaluations (see Figure 7 and Table 3).
Simple Main Effects Revealing Differences in Usage of Global Versus Local Strategies When Exploring Panels During Different Tasks (All of Them Bonferroni-Corrected).
SE = standard error.
Participants were more likely to gain insights when they explored the panels more intensely during free exploration, globally (r = .64, p = .002) as well as locally (r = .60, p = .004). This suggests that a specific explorative strategy is not relevant for having an insight although a local exploration strategy was positively linked with complexity evaluations (see Table A1 in Appendix), and a slightly negative relation between strength of insight and the average number of descriptions of texture characteristics in the think-aloud protocols (which might point to local attention) was visible as a trend: r = –.40, p = .08.
Table A1 in Appendix shows correlations between the estimated usage of explorative strategies and evaluations averaged across stimuli during haptic exploration: The more globally participants explored the panels, the higher were their average liking evaluations (for pleasingness, this positive link is a trend only); the more local the exploration, the more complex and interesting did people rate the panels. Note the potential bidirectionality of these correlations regarding causality.
Visual exploration
For the visual domain as well, complexity was a significant predictor of liking—βcomplexity = .61; F(1, 9) = 5.28, p = .05, R2adj = .300—and even more so for interest—βcomplexity = .92; F(1, 9) = 52.30, p < .001, R2adj = .837, but not so for pleasingness—βcomplexity = .10; F(1, 9) < 1, p = .77, R2adj = –.099. Note again the inhomogeneity of stimuli regarding their complexity (Figure 8). This time, the positive effect of complexity on interest was visible as a trend only when excluding Panel 4: βcomplexity = .55; F(1, 8) = 3.54, p = .10, R2adj = .220. The same is true for the negative link between complexity and pleasingness for the reduced stimulus set: βcomplexity = –.60; F(1, 8) = 4.39, p = .07, R2adj = .273. The effect on liking was not existent anymore: βcomplexity = –.06; F(1, 8) = .25, p = .88, R2adj = –.121.
Interest, liking, and pleasingness for visual stimuli of varying complexity. Error bars represent ±1 SEM.
Evaluations during haptic and visual exploration of a panel correlated significantly with each other with regard to complexity (r = .93, p < .001), interest (r = .90, p < .001), liking (r = .79, p = .004), and pleasingness (r = .82, p = .002). In both domains, the correlation between complexity and interest was stronger than between complexity and liking. Moreover, liking and pleasingness were significantly correlated for visual stimuli only (see Figure 9 for all correlations).
Correlation plot, adapted from a plot created in R and R-Studio using the package corrplot (Wei & Simko, 2017). Circle sizes represent strength of correlation and numbers their coefficients (negative coefficients are marked with a minus sign); gray circles represent nonsignificant correlations. Note that strength of insight and material feel were not examined during visual exploration.
Personality effects
We aggregated data across stimuli to examine person-specific effects. In contrast to our expectation, participants with higher scores in VVIQ_PA did not have any advantage with regard to strength of insight—r = –.11, p = .65. But strength of insight could be predicted by people’s ambiguity (in)tolerance as shown by a multiple linear regression model including subscales of the inventory for measuring tolerance of ambiguity: F(5, 15) = 2.99, p = .05, R2adj = .332 (see Table 4 for effects of each subscale). Meanwhile, need for closure did not predict strength of insight: F(4, 16) = .92, p = .48, R2adj = –.017.
Results of a Multiple Regression on Strength of Insight by Subscales of Ambiguity Tolerance.
Note. IMA = inventory for measuring tolerance of ambiguity.
Significant results are displayed in bold.
A multiple regression with the two subscales of NFT_autotelic and NFT_instrumental as predictors revealed that need for touch did not significantly affect ratings of pleasingness during haptic exploration: F(2, 18) = 3.7, p = .70, R2adj = –.068. People with high autotelic NFT were talking more during the free exploration (r = .44, p = .04), but NFT was not related to the average number of specific descriptions of texture characteristics during free exploration (rautotelic = .16, p = .49; rinstrumental = .16, p = .50). As hypothesized, persons with high need for closure explored significantly more globally on average over all tasks (r = .52, p = .02); a negative relation with ambiguity tolerance was not significant (r = –.26, p = .25; for an overview on other relations with explorative strategies see Table A2 in Appendix). We furthermore found negative relations between NFT and VVIQ_PA: For people with high need for touch, the visual vividness tends to be low; this was most pronounced in case of instrumental NFT and facial visual imagination (see Table A3 in Appendix). VVIQ_PA was furthermore negatively linked to haptic pleasingness (r = –.44, p = .05).
Discussion
We tested the Aesthetic Aha effect (Muth & Carbon, 2013) in the haptic domain and found a positive relationship between strength of insight and pleasingness under consideration of the material feel. Meanwhile, liking could be significantly predicted only if complexity was high as well in a combined model. The prediction by insight and complexity is in accordance with findings in the visual domain: Here, the Aesthetic Aha effect was stronger in contexts of perceptual challenge (Muth et al., 2016). Still, we need to consider that the present findings considering complexity might be biased by the inhomogeneity of the panels’ complexity with one panel being of very low complexity and low liking evaluations (Panel 4).
These findings indicate that haptic perception might indeed allow for pleasurable insight as evident for the visual domain—a hint to its important but often disregarded role in providing us with a grasp of meaning. Importantly, estimations of strength of insight in the current study do not equal recognition as, for example, re-recognizing a shape after a previous identification might induce a much weaker insight. Also, people could gain insights into material quality or interpret the scene with regard to an imagined storyline (e.g., one person wrote “The jellyfishes rush away because the little fish frightens them. The little fish is on a discovery tour and fascinated by all these things,” translation by the authors). In contrast to the presentation of a standardized stimulus set in random order, we preferred the successive presentation of potentially relatable haptic panels. Such an episodic context rather resembles situations in which we use touch to explore or confirm perceptual hypotheses that are integrated in a meaningful context. Meanwhile, the storyline might have increased the relevance of detected Gestalt for liking and pleasingness when compared with a more controlled setting.
Among the multiplicity of potential affective responses during haptic exploration, we focused pleasingness, liking, and interest. This was necessary given the potential overload for participants’ capacity to concentrate if explorations would take even longer (haptic exploration took approximately 1 hour). We are therefore far from an exhaustive picture of haptic aesthetics. Meanwhile, our findings indicate that liking comprises conditions under which interest (e.g., by high complexity) as well as pleasingness (e.g., by material feel and insight) are high. Interest and pleasingness were rather differentiable and in combination with liking, they were even negative predictors of each other. A slight difference to vision could indeed lie in the relevance of insight for pleasingness. Still, a direct comparison with a visual Aesthetic Aha was not possible as we did not include strength of insight in these post hoc evaluations. To do so, we would have to reduce presentation time or increase complexity of the stimuli for another group of participants given the higher speed of visual detection.
We furthermore asked if the positive link between complexity and interest (e.g., Berlyne, 1971; Muth, Westphal-Fitch, & Carbon, 2017; Silvia, 2005; Turner & Silvia, 2006) could be replicated in the haptic domain and found that indeed, interest was higher for more complex haptic stimuli (so was liking—but note that the effect on liking was weak and driven by one panel with low complexity). Furthermore, people talked more during free exploration of a panel evaluated to be interesting and this appeared to be mediated by complexity. The strong relation between complexity and interest was also evident from evaluations during visual exploration. Note that it is well possible that the inhomogeneity of stimulus complexity caused a bias.
All affective dimensions correlated positively between both modalities and when examining the correlations for each domain, we can see that liking and pleasingness are connected stronger in the visual domain. This finding is to be taken with care though because the concept of pleasingness might have a different connotation in the haptic context. It could be more strongly connected to texture characteristics of the materials—the material feel (see also decreased global exploration during pleasingness evaluation). We assume that such a focus is less clear for the visual domain and therefore pleasingness might include more diverse features that are also relevant for liking. Comparisons are furthermore difficult given the intermodal connections: It is for instance possible that people also judged simulated visual impressions of what they felt haptically and simulated or remembered haptic impressions of the textures and forms they saw. The latter influence is especially likely and problematic given the fixed order of an initial haptic and subsequent visual exploration of corresponding shapes. Indeed, some participants reported to recognize motives visually which they had been exploring haptically before. However, we were most interested in evoking insights during haptic exploration and therefore did not counterbalance the order to prevent recognition of motives that were visually perceived before.
We were additionally interested in a qualitative characterization of explorative behavior. Characteristics can only be reported for the haptic domain at this point. In subsequent research, a more reliable analysis might be gained by automated position tracking of the fingers (as suggested by an anonymous reviewer of an earlier version of this article). Furthermore, eye tracking might be useful to gain knowledge about equivalence in the visual modality. We found differences in the explorative behavior for different tasks (but note the inhomogeneity of sample sizes): Pleasingness evaluations were drawn from less global exploration than those of interest and complexity which is in accordance with the higher relevance of texture characteristics for this evaluative dimension. In contrast to our expectation, there was no advantage of global versus local exploration for having an insight; amount of exploration as such seemed relevant. There was only a trend for stronger insights when texture characteristics were described less often. We should note that local motion can comprise two explorative procedures: Local motion is not only required for extracting texture, but also for detailed contour examination which might aid shape detection or reassurance of its previous identification by global motion (see specifics of exploratory patterns and differences between global and exact shape; Lederman & Klatzky, 1993). Future studies should therefore differentiate if local exploration is focused on contour or texture. Further insights from explorative behavior are peripheral to our aims here but provide a potential direction of examination: Complexity and interest were estimated to be higher when exploration was more local and liking in contrast was higher if exploration was more global. It is important to note that these relations could be twofold: The drive to explore more globally could be a consequence of high liking (“This is enjoyable, let’s explore it all”) or—in contrast—the wider movements allow for more enjoyable experiences due to the increased activity itself. Interest could refer to refined exploration of local characteristics (“lets have a closer look, there could be finer details to detect,” see also the increase of differentiation when giving people the oportunity to manipulate a simple Gestalt by drawing, Hubbell, 1940). Alternatively, more local exploration might increase detail-detection and therefore the number of differentiable elements (complexity) which again affects interest. Properties of the materials (like roughness, flexibility, etc.) are very likely to have influenced the choice of exploration strategies. And they probably affected the ease of object recognition as well (due to higher salience of some properties and contrasts to their surroundings, see also Plaisier et al., 2008). A more controlled and systematic selection of textures and materials in future studies would be necessary to provide a corresponding analysis. However, since we aimed at generating insights by presenting a haptic storyline, we accepted possible limitations by the material properties.
Person-related effects were found for increased strength of insight when ambiguity tolerance was high, especially with regard to openness for experience; none of our hypotheses referred to ambiguity tolerance regarding people’s image of the parents and currently, we cannot explain its unexpected effect. Meanwhile, need for closure did not predict strength of insight significantly. Contrary to our expectation and to evidence from raised line drawings (using another questionnaire, though; Lebaz et al., 2012), visual vividness was not relevant here. Persons with high autotelic need for touch talked more during free exploration maybe pointing to increased elaboration as expected. They did not find the stimuli more pleasing, whereas visual vividness was negatively related to pleasingness. On a side note: People with high need for touch were found to show less visual vividness and vice versa, either pointing to the necessity to touch in case visual imagination is weak or pointing to different preferences for one modality, visual versus haptic. Exploration was indeed more global for people with high need for closure (even though we did not see that global exploration was beneficial to gaining insight, see earlier).
Conclusion
In the present article, we transferred the Aesthetic Aha effect—known for vision—to another modality. Our findings suggest that also during haptic exploration, a dynamic sense making within a semantically unstable context might lead to increased positive affect. Complexity might hereby serve as a major drive for interest. Whereas strength of insight and pleasingness were positively related when considering material feel, liking was significantly predicted only when complexity was considered as well. We need to be precautious here given the small set of stimuli and their inhomogeneity but this finding might point once more to the idea that it might not be simplicity that leads to increased liking in each case and that pleasingness differs greatly from interest. Meanwhile they successfully predicted liking in a combined model. Exploration strategies differed depending on the actual goals of the perceiver with pleasingness implying less global exploration and people’s need for closure evoking more global exploration. Gestalt detection was not related to one single strategy but to people’s tolerance for ambiguity. In sum, we can state that haptic perception plays a part in processes of understanding and can induce pleasurable insights. Further research might reveal specifics regarding the quality of this pleasure for different sensual modalities and their interplay.
It is yet too early to make a clear statement but the evidence of positive affective responses to insight by visual (e.g., Muth & Carbon, 2013), intellectual (Muth, Hesslinger, et al., 2015), and now haptic exploration makes us speculate that the Aesthetic Aha effect might reflect a very general affective response to processes of sense making. This would drive the exploration of novel situations, evoke curiosity, and motivate us to detect meaningful structure. Whereas this is a crucial precondition for learning and surviving, it would also be of relevance for the appeal of art, design, pieces of music, and haptic stimuli that defy perceptual habits but still allow for an active exploration motivated by anticipation of and rewarded by generation of insight—also those which never provide determinate solutions.
Footnotes
Acknowledgements
The authors would like to thank Julia Liss for the creation of the haptic panels and Vivien Vorndran, Ann-Kathrin Schimek, Antonia Wächter, Stefan Breitschaft, and Tanja Christiane Hansen for observation protocols as well as for help with data assessment.
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was supported by a mobility grant for the Republic of Serbia by the Bavarian Academic Center for Central, Eastern and Southeastern Europe (BayHost) dedicated to CCC.