The Interpretation of E-Motions in Faces and Bodies Derived from Static Artworks by Individuals with High Functioning Autistic Spectrum

Round 1

Reviewer 1 Report

The paper deals with a very interesting topic. The research was done methodologically correct, and the findings were interpreted in accordance with existing hypotheses. Also, this study brings some new findings that shed additional light on the status of sensitivity to emotional expression in people with ASD.

Nevertheless, in the text itself I find room for some improvements and clarifications:

General comment regarding terminology

“Bodiless heads” and “headless bodies” sounds a little scary (!), but I don't have a better suggestion for such a short description of the stimulus. Instead, expressions such as "body images with no visible head" are too complicated!

Participants

ln 168. Why is only the IQ given for the ASD and not for the TD group as well?

Stimuli

The artworks were excellently chosen as stimuli, but the authors should emphasize why they decided to use paintings, and not simpler models of faces and bodies with different emotional expressions. This is obviously a study in which different fields are connected (perception, emotions, pathology, psychological aesthetics), so it would be stressed a little more.

Procedure

Ln. 236-237. It would be good to more clearly motivate the introduction of two tasks. For example, the former is expected to provoke reproduction and the latter to recognition or something like that.

Results

Ln 389. Which post-hoc test was used?

Author Response

The authors are grateful to the three reviewers for their helpful suggestions. Below are listed our replies to their comments.

 

Reviewer 1

 

We thank the Reviewer for the helpful suggestions s/he gave us.

 

R1 <<General comment regarding terminology

Bodiless heads” and “headless bodies” sounds a little scary (!), but I don't have a better suggestion for such a short description of the stimulus. Instead, expressions such as "body images with no visible head" are too complicated!

A<<The reviewer is right (and made us smile). We used this terminology already in Actis-Grosso & Zavagno (2015). We were thinking about a possible different way of referring to our stimuli, but in the end “bodiless heads” and “headless bodies” were the best we can do. Thus, given that this paper is a “replication” (with ASD) of Actis-Grosso & Zavagno (2015), we prefer to maintain this terminology.

R1 << ln 168. Why is only the IQ given for the ASD and not for the TD group as well?

A<<This a typical procedure in the literature with ASD, in order to divide high functioning (i.e. former defined as Asperger) ASD individuals from low functioning ones. If the former is the case, then the control group is formed by typically developed individuals (as in our study), whereas low functioning ASDs are usually compared with low IQ groups of individuals (such as, for example, individuals with the Down Syndrome). For this reason it is usual to have, together with the diagnosis of ASD, a reliable measure of the IQ as well. This is not true for typically developed population, and we do not think that our ethical committee would allow us to measure the IQ of our students, which, however, should be in the range of 90-120 (hopefully…)

R1 << The artworks were excellently chosen as stimuli, but the authors should emphasize why they decided to use paintings, and not simpler models of faces and bodies with different emotional expressions. This is obviously a study in which different fields are connected (perception, emotions, pathology, psychological aesthetics), so it would be stressed a little more.

A<< We are not sure about this comment. The aim of our study, as repeatedly stated throughout the paper, is focused on the visual arts, thus it is obvious that we are testing paintings instead of “simpler models of faces and bodies with different emotional expressions”. At p. 26 we explicitly stated “The present study is the first attempt to test how ASDs interpret visual artworks in terms of their emotional content”. Furthermore, it is stated at p. 3 that “Emotion processing and emotion recognition have been widely investigated in ASD individuals” and a paragraph of the Introduction is dedicated to show main results from these studies. Thus “simpler models of faces and bodies with different emotional expressions” were already extensively studied by other researchers, while nobody has investigated artistic representation of emotions, which is the aim of our study.

R1 << Ln. 236-237. It would be good to more clearly motivate the introduction of two tasks. For example, the former is expected to provoke reproduction and the latter to recognition or something like that.

A<< We think that the reviewer is not referring to the two tasks, but to the two options (i.e. option A and option B) for task 1. We also think that the reviewer refers to the difference between recognition and recall, which of course could be taken into consideration when asking a participant to answer an open (i.e. recall) vs. a close (i.e. recognition) question. However, in our case with the “open question” we were asking participants to freely identify the affective state that they thought was depicted on the visual artwork, which cannot be considered a “recall”. If they were unable to do it, the experimenter was “suggesting” possible emotions, and, again, we do not think this could be considered “recognition” (this point is addressed in the Discussion section). Given that the only reason for having two options was to help ASDs in giving an answer, we decided to leave the description of the two options as it was in the former version of the manuscript.

R1 << Ln 389. Which post-hoc test was used?

A<< We are very sorry, but we do not understand this question. We are always referring to “t-test post-hoc” and of course no post-hoc were performed on chi squares.

Reviewer 2 Report

The manuscript titled ‘The interpretation of e-motions in static artworks by individuals with high functioning autistic spectrum’ compared perceived emotions and dynamicity associated with artistic representations of faces and bodies in two groups of participants—individuals with Autism Spectrum Disorders (ASD) and typically developed individuals (TD). The study found an interesting pattern of results, showing that ASD participants did not differ from TD ones in the ability of attributing emotional states to artistic representations of faces and bodies, but were more likely to attribute negative emotions to stimuli interpreted as neutral by TD participants.

Overall, I enjoyed reading this work: it is an original and important research, and I believe it has the potential for being a significant contribution to knowledge. However, in its current form the manuscript is unclear about or does not report key information about data analysis that will enable its interpretation and replication. For example, results from the ratings for perceived intensity of emotions are described in the methods section, but they are not analysed nor discussed later in the manuscript. I think these results will be quite interesting, so would suggest including them in the paper. Otherwise, I would suggest for the authors to remove them entirely from the list of collected measures. Below I share my comments, that I hope will help improving the manuscript.

I have one main concern about the title, which I think can be misleading in its current form. As the authors acknowledge in the discussion, results from this study cannot be generalised to the ability of identify emotions –also because we do not know which emotion each artist intended to convey in their artworks. Similarly, I believe that the results from this study cannot be generalised to artworks more in general. I argue that the visual manipulation of the stimuli used in this study resulted into a very different image compared to the composition and richness of the original artworks—some of the stimuli are indeed only details of a much bigger painting. For this reason, I would suggest changing the title into something like: ‘The interpretation of e-motions in artistic representations of faces and bodies by individuals with high functioning autistic spectrum’, which I think better describes the authors’ work.

I do not think the authors can claim that Michelangelo’s studies were ‘conceived by the artists as “non emotional” ‘ (page 6, line 179), unless have historical sources referring to these specific drawings. It is equally plausible that the artist made a series of sketches to study a particular emotion, or indeed neutral emotions. I would suggest it is more appropriate to talk about the emotions that are most frequently associated with a particular stimulus, also based on the previous study by Actis-Grosso and Zavagno (2015).

The authors interpret the lack of reporting neutral emotional states in the ASD participants as a possible effect of their training (line 516). It would be interesting if the authors suggested how to address this issue, either with possible alternative guidelines for training people with ASD or by experimentally testing this hypothesis (e.g., individuals with ASD that did not receive this specific training).

Page 25, line 629: I do not think the authors can claim that ASD participants had ‘a higher ability in recognizing that faces and bodies are authored by different artists’ as they did not explicitly ask participants this information, but they infer it from the dynamicity measure.

Page 10, line 275, Results section: the text refers to a ‘list B’, but the authors do not provide a definition of such list. The authors later refer to list B as containing the ‘different emotions chosen by the two groups’ (line 347). This definition should be specified earlier and the difference between list A and B should be clarified.

Why the chi-squares reported in lines 286 and 288 have four degrees of freedom? By reading the information reported by the authors, I am assuming that the analysis had three columns (positive, negative and neutral emotions) and two rows (ASD and TD), so the degrees of freedom should be 2.

Across the paper, the authors perform repeated chi-squares. For example: in section 3.1.2 (pp10), they do three chi-squares on the frequency of use of the most used words respectively for positive, negative and neutral emotions (line 302 to 308). My understanding is that these frequencies are extracted from different sections of the same experiment; therefore, I would suggest adjusting the p value for repeated testing (e.g. in this case should be .05/3= .016, so the authors should accept as significant only tests that have p < .16). Similarly, in the section 3.1.4 my understanding is that the authors performed a fourteen chi-square tests, one for each stimulus—is my interpretation correct?

Section 3.1.5 does not clearly report on which type of frequencies the chi-square was performed on. This information should be specified.

Section 3.2 should specify the outcome variable on which the ANOVA was performed. I am assuming the outcome variable were the dynamicity ratings, but I would recommend indicating it clearly in the first paragraph of the text, separating the ANOVA model from its main results.

Some minor comments:

Page 3, line 71: change ‘confirm’ with ‘confirmed’?

Page 3, line 88: Do the authors mean ‘visual input’?

Page 6, section 2.1: were ASD participants rewarded?

Page 20, line 482: ‘challenge’ –rather than ‘shape’—feels more appropriate in this context.

Figures 8 and 9 could be improved by reporting the percentages associated with each emotion, across the two groups.

 

Author Response

Reviewer 2

We thank the Reviewer for the helpful suggestions, which helped us improve in clarity the manuscript. Following are the answers to her/his concerns (sometimes extrapolated from a larger discussion).

 

R2 << For example, results from the ratings for perceived intensity of emotions are described in the methods section, but they are not analysed nor discussed later in the manuscript.

A<< The reviewer is right and we apologize for this mistake, which incredibly went unnoticed not only by the three authors but also by two reviewers out of three (Reviewer 2 is the only one who properly noticed this). The truth is that, at a first stage of the study, we intended to analyse those data as well, but after realizing (i) that ASDs had too many missing values and (ii) that the paper was already quite long, we decided not to further analyse those data, the interpretation of which would require a quite long (and maybe too speculative) discussion (and also in light of some methodological concerns). This is now explicitly stated in the Section entitled “Preliminary classification of emotion” with the following sentence “After this analysis, we decided not to analyze (at least not at this stage) data obtained with Task 2 (i.e. ratings for perceived intensity of emotions), given the high number of missing values from ASDs in list A (as better detailed below), which raises some methodological concerns about the legitimacy of such an analysis (and consequent interpretation).”

R2 << I have one main concern about the title, which I think can be misleading in its current form. As the authors acknowledge in the discussion, results from this study cannot be generalised to the ability of identify emotions –also because we do not know which emotion each artist intended to convey in their artworks. Similarly, I believe that the results from this study cannot be generalised to artworks more in general. I argue that the visual manipulation of the stimuli used in this study resulted into a very different image compared to the composition and richness of the original artworks—some of the stimuli are indeed only details of a much bigger painting. For this reason, I would suggest changing the title into something like: ‘The interpretation of e-motions in artistic representations of faces and bodies by individuals with high functioning autistic spectrum’, which I think better describes the authors’ work.

A<< We agree with the Reviewer, especially with reference to the generalization of the present study to artworks more in general. Indeed, we acknowledge this point in Actis Grosso & Zavagno (2015) and in this manuscript we report the “contextual effect” by saying “our hypothesis was ultimately based on a sort of “contextual effect”, where the emotions conveyed by the face may either enhance or contrast the emotion conveyed by the body, and the final dynamic/emotional content of the whole portrayed figure may in turn either enhance or contrast the global emotional and dynamical impression of the whole artwork”. Next steps of our project are related to studies on “much bigger paintings”. This said, we thank the Reviewer once again. The title has been changed into “The interpretation of e-motions in faces and bodies derived from static artworks by individuals with high functioning autistic spectrum”.

R2 << I do not think the authors can claim that Michelangelo’s studies were ‘conceived by the artists as “non emotional” ‘ (page 6, line 179), unless have historical sources referring to these specific drawings. It is equally plausible that the artist made a series of sketches to study a particular emotion, or indeed neutral emotions. I would suggest it is more appropriate to talk about the emotions that are most frequently associated with a particular stimulus, also based on the previous study by Actis-Grosso and Zavagno (2015).

A<< “non emotional” has been deleted.

R2 << The authors interpret the lack of reporting neutral emotional states in the ASD participants as a possible effect of their training (line 516). It would be interesting if the authors suggested how to address this issue, either with possible alternative guidelines for training people with ASD or by experimentally testing this hypothesis (e.g., individuals with ASD that did not receive this specific training).

A<< Unfortunately (or fortunately, depending on one’s perspective) it is almost impossible to find individuals who, being diagnosed as ASD, have not been trained in recognizing emotional faces (usually they are trained in the six “basic” emotions and possibly on the subjective feeling related to them). We do not think at this stage of our project to be allowed to suggest guidelines for training people with ASD, even though we do think that it would be convenient to train them for emotional bodies as well. However, we think our data are too preliminary to suggest anything to educators who are working with ASDs and, as said, we know that this issue is experimentally unavoidable.

R2 << I do not think the authors can claim that ASD participants had ‘a higher ability in recognizing that faces and bodies are authored by different artists’ as they did not explicitly ask participants this information, but they infer it from the dynamicity measure.

A<< The proposition “which in some cases is translated in a higher ability in recognizing that faces and bodies are authored by different artists,” has been deleted

R2<< Results section: the text refers to a ‘list B’, but the authors do not provide a definition of such list. The authors later refer to list B as containing the ‘different emotions chosen by the two groups’ (line 347). This definition should be specified earlier and the difference between list A and B should be clarified.

A<< In the revised version of the manuscript list B is hopefully better explained.

R2<< Why the chi-squares reported in lines 286 and 288 have four degrees of freedom? By reading the information reported by the authors, I am assuming that the analysis had three columns (positive, negative and neutral emotions) and two rows (ASD and TD), so the degrees of freedom should be 2.

A<< The degrees of freedom have been corrected, with thanks.

R2<< Across the paper, the authors perform repeated chi-squares. For example: in section 3.1.2 (pp10), they do three chi-squares on the frequency of use of the most used words respectively for positive, negative and neutral emotions (line 302 to 308). My understanding is that these frequencies are extracted from different sections of the same experiment; therefore, I would suggest adjusting the p value for repeated testing (e.g. in this case should be .05/3= .016, so the authors should accept as significant only tests that have p < .16).

A<< This is not what we actually did. Instead, what we did was to consider only those words that have been used at least five times by the two groups of participants. These words were three (i.e. serenity, joy and happiness) for positive words and 7 (i.e. fear, scare, anger, sadness, disgust, pain and despair) for negative words, as shown in the graphs. We considered each group of words as independent, and that’s why the degrees of freedom in the first case are 2 (i.e. three columns, one for each word, and two rows, TD and ASD) and in the second are 6 (i.e 7 columns). The Reviewer should consider that the “division” of words in positive, negative and neutral is a matter of choice (which means somehow arbitrary), because we could decide as well to divide words only in positive and negative, for example (putting half neutral in positive and half in negative), or to divide them in 5 categories, such as very positive, positive, neutral, negative, very negative…In the first case we could divide the p value by 2, in the second by five? Given that words were spontaneously spoken by our participants, we did not have the choice to categorize them in advance. The same holds for the choice we made to consider only those words that have been used at least 5 times: we could as well consider the words used three times (or two, or one hundred…). We are not saying that this is the “right” way to proceed, and we are very open to other suggestions, but we think that to divide the p value by three would be as arbitrary as to divide it by 98 (which is the total number of words used by participants), or by 12 (negative+positive words used 5 times..). This said, please consider that if we divide the p value by three, nothing changes (we rounded to 0.02 exactly 0.016…).

R2<< Similarly, in the section 3.1.4 my understanding is that the authors performed a fourteen chi-square tests, one for each stimulus—is my interpretation correct?

A<< We proceeded in this case in a way similar to what is described above: for each stimulus (considered as independent) we ran a chi square on the words chosen by participants (that’s why the degrees of freedom are different in the two reported cases).

R2<< Section 3.1.5 does not clearly report on which type of frequencies the chi-square was performed on. This information should be specified.

A<< The chi square was performed on list B, this information is now added in section 3.1.5

R2<< Section 3.2 should specify the outcome variable on which the ANOVA was performed. I am assuming the outcome variable were the dynamicity ratings, but I would recommend indicating it clearly in the first paragraph of the text, separating the ANOVA model from its main results.

A<< We are not sure about this comment. Section 3.2. is entitled “Dynamicity ratings” (already in the previous section), then when the reviewer says “I am assuming the outcome variable were the dynamicity ratings”, we think that maybe something is missing…By the way, hoping not to be redundant, we also added this information in the text.

R2<< Some minor comments:

Page 3, line 71: change ‘confirm’ with ‘confirmed’?

A<< Done, with thanks

R2<< Page 3, line 88: Do the authors mean ‘visual input’?

A<< Yes, and the word “visual” has now been added

R2<< Page 6, section 2.1: were ASD participants rewarded?

A<< We thank the reviewer for this question, which made us realize there was an error in the description of participants. In fact ALL participants (i.e. both ASD and TD) were not rewarded, as it is now clearly stated in section 2.1

R2<< Page 20, line 482: ‘challenge’ –rather than ‘shape’—feels more appropriate in this context.

A<< Done, with thanks

R2<< Figures 8 and 9 could be improved by reporting the percentages associated with each emotion, across the two groups.

A<< Even though the reviewer might be right, we do prefer not to add the percentage in the figure but to report instead labels for different emotions. Adding percentages (in other words, having both the label and the corresponding percentage) on the figure has a chaotic outcome, which is more confounding than improving the clarity of the figure itself.

Reviewer 3 Report

The paper is concerned with the perception and recognition of emotions in faces and body postures in static artworks. It is assumed that some emotions, such as anger and fear, are more related to motion than other emotions, such as puzzlement. The authors call the former emotions ‘E-motions’ and hypothesized that they, even if expressed in static representations of faces or bodies, convey more dynamics than the other emotions. They were also interested in whether individuals with Autism Spectrum Disorders (ASDs) are impaired in recognizing the dynamic-related emotions, compared to typically Developed (TD) individuals. In their study they required participants from both groups to first identify the emotions in presented pictures of bodiless heads and headless bodies and then to rate the emotions and dynamicity of these objects. One of several results was that individuals with ASD identified negative emotions more frequently than TD individuals, and TD individuals identified neutral emotions more frequently than ASD participants. This suggests that individuals with ASDs perceive the visual artworks in a different way than TD individuals. This and the other implication of the results are discussed.

The topic of the study is relevant and interesting, and the experiment is conducted professionally. The paper is also well written. I did not detect any serious errors or weaknesses and think that the paper provides a valuable contribution to the field of emotion perception and aesthetics. Merely the minor point below should be addressed before publication.

Minor

Figure 2. The bars for neutral affective state have lengths that differ from those stated in the text (reversal). 

Author Response

Reviewer 3

We thank the Reviewer, especially because s/he was the only one to notice the error in the graph, for which we apologize. The error has been corrected, with many thanks!