Through visiting museums, we are able to come into contact with things not usually seen in our daily lives, such as rare creatures, historical relics, archaeological remains, and artwork. Through the appreciations of museum exhibits, we can experience virtual lives in the spatial and temporal contexts that originally encompassed the exhibited items with vivid reality. Great progress in recent technologies related to virtual and augmented reality has allowed museums to incorporate three-dimensional (3D) representation into their exhibits (Walczak and White, 2003; Hirose, 2006; Petridis et al., 2006). Simultaneously, multimodal displays based on similar technology have been equipped to appeal to not only visual but also auditory, haptic, and olfactory modalities (Butler and Neave, 2008; Figueroa et al., 2009). Such new methods contribute to the diversity of presentation within and between museums, comprising a new category of “virtual museum” that distributes digital replicas of exhibits to each individual, going beyond the conventional static presentations that remain inside glass cases.

Multimodal displays are likely to facilitate a deeper understanding of museum exhibits. Indeed, it has been found that when haptic devices are incorporated into exhibits, visitors take more time to appreciate them (Butler and Neave, 2008); moreover, the inclusion of haptic devices in exhibits has received positive evaluations from not only visitors, but also museum curators and researchers (Asano et al., 2005). Thus, given that one of the objectives of museums is to educate visitors about their exhibits, it is important to improve visitors' impressions of exhibits and to encourage longer appreciation times by adding haptic information.

Directly touching museum exhibits can provide visitors with additional information on the texture, weight, and materials of items than merely viewing them can. In cognitive psychology, it has been suggested that the provision of more information about unknown things can elicit a positive affect (Biederman and Vessel, 2006; Yamada et al., 2012, 2013). These findings may underlie why the addition of haptic information to exhibits leads to more positive impressions of them. However, there is a dilemma in introducing such touchable museum exhibits. While exhibits have traditionally been susceptible to age-related natural deterioration and damage during transport, which can affect the condition of exhibit items to varying degrees, the touching of exhibit items by visitors' hands can also accelerate their deterioration.

To overcome this dilemma, in our study we focused on the addition of a simple object, a box, that has the same weight to exhibits. This is information for which “surrogates” (i.e., information that does not require the direct touching of original exhibit items) can be easily created and that does not seem to be directly related to positive or negative emotions regarding the exhibit. Moreover, to investigate such an addition experimentally, we can easily prepare and control the experimental stimuli when compared with texture and material stimuli. Humans perceive the weight of a stimulus based on outputs from proprioceptive and cutaneous sensors that constitute haptic perception (Jones, 1986; Flanagan et al., 1995; Flanagan and Wing, 1997). Moreover, many previous studies with psychophysics have revealed that haptic perception sometimes alters the appearance of visual stimuli (Ernst et al., 2000; Violentyev et al., 2005; Ide and Hidaka, 2013). Thus, we developed the hypothesis that weight cues will provide a portion of the available haptic information about museum exhibit items without direct touch, thereby resulting in positive effects on visitors' visual museum appreciation.

The goal of the present study was to examine whether weight cues affect visitor appreciation of museum exhibits. In our experiment, we prepared polystyrene foam boxes containing sand in accordance with the actual weight of each exhibit. We divided the participants into two conditions; in one, the participants had the opportunity to pick up the weight stimuli using their hands (cued condition), and in the other the participants had no opportunity to do so (uncued condition). We predicted that if our hypothesis was correct, that is, that weight cues affect participants' appreciation of exhibit items, then the participants in the cued condition would (1) have significantly more positive perceptions of the exhibit appreciation experience and of the museum itself, (2) be able to better recall a greater number of the exhibits, (3) look at exhibits for longer periods of time, and (4) be willing to pay more money to experience such appreciation when compared with the participants in the uncued condition.

A post-experiment interview showed that one female participant had previously visited the museum, so data from her were not used for analysis. We compared the recognition performance, viewing duration, and willingness-to-pay (WTP) between the cued and uncued conditions. Recognition performance was indicated by A' (Figure 2A) and B”D measures (Donaldson, 1992). Viewing duration of each exhibit was calculated by dividing total duration by the number of exhibits (i.e., 19) and was analyzed after log-transformation (Figure 2B).

For the recognition performance, a mixed analysis of variance of the A' measure with cue condition as a between-subjects factor and box condition as a within-subjects factor was performed; results showed that only a main effect of cue was significant, F_{(1, 39)} = 5.31, p < 0.03, η²_p = 0.14. Although there was no interaction between cue and box conditions, we individually conducted paired t-tests and showed that the difference between the box-present and box-absent conditions was significant for the participants in the cued condition, t₍₁₈₎ = 2.32, p < 0.04, Cohen's d = 0.82, while no significant difference between the conditions was found for the participants in the uncued condition, t₍₂₁₎ = 0.29, p > 0.77, Cohen's d = 0.08. A two-sample t-test showed that the difference between the cued and uncued condition was significant for the box-present condition, t₍₃₉₎ = 2.60, p < 0.02, Cohen's d = 0.85. On the other hand, ANOVA and individual t-tests did not show any significant effect in the B”D measure.

As for the viewing duration, a two-sample t-test revealed that participants in the cued condition viewed the exhibits significantly longer than participants in the uncued condition, t₍₃₉₎ = 2.50, p < 0.02, Cohen's d = 0.80.

As for WTP, a two-sample t-test showed no significant difference between the cued and uncued conditions, t₍₃₉₎ = 1.53, p > 0.13, Cohen's d = 0.49 (Figure 2C).

To investigate the relationship between the indices analyzed above and participants' impression of their exhibit experience as well as the museum, we used an exploratory factor analysis on the individual questionnaire data that extracted principal components with the unweighted least-square method and promax rotation after reversed items were adjusted. Bartlett's test of sphericity showed significant χ² values [exhibit appreciation scale: χ²₍₉₁₎ = 423.73, p < 0.0001; museum scale: χ²₍₅₅₎ = 163.54, p < 0.0001]. The Kaiser-Meyer-Olkin measurement of sampling adequacy was equal to 0.78 and 0.67 for the exhibit appreciation and museum scales, respectively, suggesting that these data were suitable for factor analysis. Curves in the scree plot suggested that two factor solutions could be extracted from both the scales (Tables 1, 2). These factors explained 59.2% and 48.1% of the total variances in the exhibit appreciation and museum scales, respectively. For the participants' exhibit appreciation, the first and second factors included items on “likability” and “contentment” with their exhibit appreciation experience, respectively. For the impressions of the museum, the first and second factors included items on the “value” and “quality” of the museum, respectively.

Moreover, using each of the indices (recognition performance, viewing duration, and WTP) and the four factor scores, we performed a path analysis (Figure 3). The goodness of fit of a model to the data was high, χ²₍₁₅₎ = 24.95, p > 05; RMR = 0.038; GFI = 0.856; AGFI = 0.731. We estimated 95% confidence intervals using a Bayesian framework with 100,000 iterations after a burn-in of 20,000 iterations. Through the resampling, all variables converged under the convergence criteria of 1.002 (Gelman et al., 2004). An indirect effect of the viewing duration on recognition performance and WTP was significant, 95% confidence intervals of the standardized indirect effects were 0.019 to 0.222 and 0.047 to 0.246, respectively.

Furthermore, we compared the four factor scores between the cued and uncued conditions. The results showed that there was no difference, suggesting that the manipulation of the weight cue did not directly affect the participants' impression.

Finally, since one can argue that any gender difference might have impacted on our results, we ran male vs. female comparisons with two-sample t-tests for viewing duration, recognition performance (A' and B”D values), and WTP. The results did not show any significant differences between males and females for viewing duration: t₍₃₉₎ = 1.29, p > 0.20, Cohen's d = 0.45; A': t₍₃₉₎ = 0.69, p > 0.49, Cohen's d = 0.24; B”D: t₍₃₉₎ = 0.23, p > 0.81, Cohen's d = 0.08; nor WTP: t₍₃₉₎ = 1.35, p > 0.18, Cohen's d = 0.47, and hence, the unexpected effect of gender on our results was foreclosed.

The results of the mean differences between the groups based on the presence of weight cueing suggest significant effects of cue for each behavioral index. First, participants who were instructed to lift the weights showed significantly higher memory performance relative to the other participants. Although this was not confirmed under the significant interaction, the cued participant's memory performance of the box-present exhibits was higher than that of the box-absent exhibits. This effect was not due to a gender difference. Moreover, considering all the participants viewed the boxes, the explanation that the box merely served as a visual marker for memory retrieval does not seem plausible. Instead, the results might come from the instruction itself in a manner of increasing the participants' arousal or motivation to the appreciation.

Second, consistent with the results of a previous study (Butler and Neave, 2008), our results showed that participants with access to haptic experience took a significantly longer time to appreciate exhibits than participants without such haptic experience access. One could argue that this time difference reflects the time spent on the lifting action itself. The mean difference of the total viewing duration between groups was approximately 106.6 s and the total duration required for lifting action that was measured in a supplementary experiment was approximately 19.7 s (see Appendix). Even when this lifting duration was subtracted from the total viewing duration in the cued condition, the significant difference between the cued and uncued condition in viewing duration per exhibit was preserved, t₍₃₉₎ = 2.26, p < 0.04, Cohen's d = 0.71. Therefore, it is unlikely that merely the lifting action of four boxes produced the difference observed in the main experiment. Third, the difference in the WTP variable 102.99 Japanese yen higher in the cued condition did not reach significance, suggesting that the weight cue alone was not sufficient to significantly change the participants' WTP directly.

Analyses of the questionnaire data provided much evidence supporting the group differences. The factor analysis showed that the questionnaire items could be categorized into two principal factors: the factors “likability” and “contentment” with respect to the participants' impressions of the exhibit appreciation experience, and factors “value” and “quality” with respect to the participants' impressions of the museum itself. The results of the path analysis suggest that the formation of participants' impressions regarding viewing the museum exhibits was influenced by likability-based processing regarding exhibit appreciation and quality-based processing regarding the museum itself after establishing the impression of value. One implication of these results is that curators' efforts to facilitate the visitors' memories of exhibits may not necessarily lead to their positive attitudes toward paying more for similar museum experiences. Our findings of no significant correlation between WTP and recognition performance, r = 0.02, p > 0.91, support this implication.

How did the weight cues affect the results? In light of the internal mechanisms of participants, as mentioned in the introduction, we first surmised that the greater amount of information provided from multiple modalities may lead to higher processing fluency for the corresponding objects, thereby enhancing the esthetic pleasure experienced through perceiving those objects (Reber et al., 1998, 2004; Kuchinke et al., 2009). However, this might not be true or at lease it is somewhat premature to draw a conclusion on this issue, because the results showed that the manipulation of the weight cue did not directly affect any factor scores. The results of the path analysis instead indicate another possibility. We just found that the weight cue significantly prolonged the viewing duration and a model with the viewing duration as a causal factor could explain the results. Hence, an indirect effect of the manipulation of the weight cue might exist. That is, the lifting action prolongs the viewing duration, and then the viewing duration changes participants' impression of exhibits and museum (for example, a significant direct effect of viewing duration on value was found), and then these impression changes affect recognition performance and WTP through separate paths. These findings have important implication on a future museum exhibition: Interaction with even a box with a neutral appearance does change visitors' impression, memory, and value estimation for exhibits.

In conclusion, both the data for each behavioral index and the questionnaire data indicate that the interaction with physical surrogate objects is (at least indirectly) sufficient to facilitate visitors' appreciation of museum exhibits. Future studies should address to what degree the weight information is important. For example, one could hypothesize that weight information does not contribute to the appreciation of pictorial art exhibit items because the esthetic value of such paintings is obviously unrelated to their physical weight. Furthermore, it is unclear whether weight information itself is important. Although visitors do not know the actual weight of the exhibits, visitors can also guess somewhat the weight of the exhibits on the basis of the visually perceived size and texture of them. Hence, a loose matching between the weight cues and the exhibits might matter. Moreover, depending on the shape of the exhibits, there is a difference in the weight distribution between the weight cues and the exhibits. How these issues influence the visitors' appreciation should be clarified in future. Owing to the recent progress in 3D printing technology, 3D printers can now easily produce precise replicas of all kinds of items for application in museum exhibits (Allard et al., 2005; Kelley et al., 2007; Niven et al., 2009). Such surrogate exhibits allow visitors to directly touch items freely without having to worry about damaging the original items. The 3D replica-based experiments would resolve the issues about weight information we discussed above. Further cross-disciplinary investigations between museology and psychology that focus on the role of such surrogate exhibit items are needed for a deeper understanding of not only the mental mechanisms involved in the appreciation of museum exhibits, but also their creation and presentation.