Does it matter how you ask? Self-reported emotions to depictions of need-of-help and social context

Background When humans observe other people’s emotions they not only can relate but also experience similar affective states. This capability is seen as a precondition for helping and other prosocial behaviors. Our study aims to quantify the influence of help-related picture content on subjectively experienced affect. It also assesses the impact of different scales on the way people rate their emotional state. Methods The participants (N=242) of this study were shown stimuli with help-related content. In the first subset, half the drawings depicted a child or a bird needing help to reach a simple goal. The other drawings depicted situations where the goal was achieved. The second subset showed adults either actively helping a child or as passive bystanders. We created control conditions by including pictures of the adults on their own. Participants were asked to report their affective responses to the stimuli using two types of 9-point scales. For one half of the pictures, scales of arousal (calm to excited) and of bipolar valence (unhappy to happy) were employed; for the other half, unipolar scales of pleasantness and unpleasantness (strong to absent) were used. Results Even non-dramatic depictions of simple need-of-help situations were rated systematically lower in valence, higher in arousal, less pleasant and more unpleasant than corresponding pictures with the child or bird not needing help. The presence of a child and adult together increased pleasantness ratings compared to pictures in which they were depicted alone. Arousal was lower for pictures showing only an adult than for those including a child. Depictions of active helping were rated similarly to pictures showing a passive adult bystander, when the need-of-help was resolved. Aggregated unipolar pleasantness and unpleasantness ratings accounted well for arousal and even better for bipolar valence ratings and for content effects on them. Conclusion This is the first study to report upon the meaningful impact of harmless need-of-help content on self-reported emotional experience. It provides the basis for further investigating the links between subjective emotional experience and active prosocial behavior. It also builds upon recent findings on the correspondence between emotional ratings on bipolar and unipolar scales. Electronic supplementary material The online version of this article (doi:10.1186/s40359-015-0066-3) contains supplementary material, which is available to authorized users.

analyses presented in the main paper ensures that the effects of other analyses were not confounded with effects stemming from the booklet itself. To provide a complete analysis, we assessed whether language, scale order or dimension order inuenced the ratings of participants. We found that none of these factors had more than a minor inuence on their ratings of the four assessed dimen- sions. An overview of all effects stemming from the differences in the rating booklets' version is provided in Table S2. The only difference between comparable booklet versions that just reached a medium effect size was that participants who found the arousal and pleasantness scales on top of each page reported a higher amount of mixed feelings, i.e. a more frequent co-occurrence of pleasant and unpleasant emotions as response to one picture. A possible explanation for this effect could be that when prompted to rate their pleasant feelings first, participants are more likely to report unpleasant ones too, whereas reporting the intensity of unpleasant feelings first seems to decrease the likelihood of also reporting pleasant feelings. Pre-defined picture content is reflected in pattern of mean ratings Distinct picture categories were created to assess emotional ratings to need-of-help depictions and social context variations, including active helping. The category specific averaged responses are illustrated in Figure S1 illustrates that a-priori dened picture categories lead to distinct emotional ratings on each type of rating scale.
Pictures showing birds or children within the same need of help category clearly cluster together according to both rating dimensions of both scale types. Pictures encompassing the "social context" subset rather form an independent cluster.

Detailed results of aggregated ratings' analyses
In the main article we have shown that mean arousal and bipolar valence ratings per picture can be expressed as aggregated pleasantness and unpleasantness ratings. Specifically, arousal ratings can be inferred from the sum of pleasantness and unpleasantness ratings, bipolar valence ratings from the difference between pleasantness and unpleasantness ratings (see Figure S2).   Table 2 and 3 of the main article) only reveals a difference in gender effects on mean ratings across pictures. Comparing the meaningful differences emerging for unpleasantness and pleasantness ratings to the ones obtained for bipolar valence (see Table 2 and 3 of the main article), the comparison between "child alone" and "social helping" pictures is the only one that leads to another interpretation of ndings. Table 3 shows the detailed results of the analyses of aggregated pleasantness and unpleasantness ratings regarding picture content's and gender effects. Results for the "need of help" subset are shown on top, whereas those for the "social context" subset are underneath.

Mixed feelings
The results reported so far have pointed out similarities between ratings made on the two scale types used. However, one conceptual difference between assessment of bipolar valence and unipolar pleasantness and unpleasantness is that co-occurrence of pleasant and unpleasant feelings, so called "mixed feelings", can only be measured on unipolar scales. As an indicator of so called mixed feelings the smaller value out of the pleasantness and unpleasantness ratings for each picture (min(pleasantness, unpleasantness)) was used(Schimmack, 2001; see Figure 2). Scores of mixed feelings were then averaged per picture. If the average intensity of mixed feelings is greater than 0, participants indicated to have some degree of pleasant and unpleasant feelings at the same time.
Across all pictures, the mean intensity of mixed feelings was about 1 rating step greater than 0, M = 1.08, [1.03, 1.14], with a narrow confidence interval indicating only little uncertainty for this finding. One explanation for this finding could be that participants avoided the extreme ends of the rating scale (Guilford, 1954). To assess this possibility, we inspected the distribution of minimal pleasantness and unpleasantness ratings and found that more than two thirds (6767 out of the 9818 ratings with valid pleasantness and unpleasantness values) were 0 or 1. Hence, there was no evidence for an aversion to use the extreme ends of the unipolar rating scales.
Rather, these results suggest that participants reported genuine co-occurrence of pleasant and unpleasant emotions, i.e. mixed feelings.
The strong linear correlation between the difference of pleasantness and unpleasantness ratings and bipolar valence (see Figure

Mixed feelings are scarcely affected by picture content
One last question that arises is whether mixed feelings -as an additional measure of subjective emotional experiences only assessable on unipolar scales -are influenced by picture content in the context of our study. Hence, picture content analyses were conducted for mixed feelings just as for the four explicit ratings. Results suggested that mixed feelings were affected by help-related picture content to a certain degree and mainly for ratings provided by women (see Table S4 and Figure S3).
For women only, pictures showing a child in need-of-help rather than no-needof-help elicited somewhat fewer mixed feelings and they also tended to exhibit less mixed feelings with regard to child compared to bird pictures. Both differences were medium in size with CIs ranging also into negligible values. The difference in mixed feelings for need-of-help compared to no-need-of-help depictions was evident for pictures of children not for such of birds (see Figure S4 A).
When assessing mixed feelings for pictures of the "social context" subset, women but not men reported more mixed feelings to "child-alone" compared to "adultalone" or "social-helping" pictures (see left side of Figure S4 B). Men tended to report more mixed feelings for "social-helping" pictures than for "adult-alone" ones (see right side of Figure S4 B). Thus, social context had no consistent effects on mixed feelings and differences for both men and women displayed large condence intervals, reecting considerable uncertainty associated with the effect. Moreover, no discernible social context category stood out for either gender, however it seems that gender differences with regard to the presence and the intensity of mixed feelings Effects of social contextual content are shown in the bottom panel with white cat's eyes representing "child-alone" pictures, light gray ones "adult-alone" pictures, middle gray ones "social-passive" and dark gray ones for "social-helping" pictures. The length of cat's eyes indicates 95% confidence intervals.
to certain content categories might be easier to uncover using unipolar pleasantness and unpleasant scales rather than valence ratings.