1. Neuroscience
Download icon

A social chemosignaling function for human handshaking

Research Article
Cite this article as: eLife 2015;4:e05154 doi: 10.7554/eLife.05154
4 figures, 4 videos and 2 additional files

Figures

Handshakes can transfer chemosignaling components.

(A) A representative image of our sampling method using a nitirle glove during handshake. (B) An example chromatogram from one experiment. Note that the ‘clean’ condition is a glove worn by the same hand, but not shaken. This controls for potential contamination from the glove-wearing hand. Most unmarked peaks in the chromatogram that are present in both the clean and the shaken are PDMS artifacts—various siloxane compounds that come from either the twister used to sample the gloves or the GC column. Moreover, some peaks that are present in this example were not present across subjects. The only three peaks that were present following all shakes but never once in control are those we describe in the following panel. (C) Summated data from 10 individuals (each an average of three shakes) demonstrating three compounds of interest in chemosignaling (Gallagher et al., 2008) that were effectively transferred by handshaking in all instances and never once in control. Error bars are standard error, **p < 0.01, ***p < 0.001.

https://doi.org/10.7554/eLife.05154.003
Figure 2 with 1 supplement
Humans often touch their own face and concurrently sniff.

(A) Agreement in scoring of 153 subjects across two independent raters. (B) Total face touching duration during the 1-min baseline. (C) Spatial distribution of face touching during the 1-min baseline. Grid reflects 17 facial regions. (D) Measure of nasal airflow during baseline vs the time when a hand was at the face. Subjects that increased flow in blue and subjects that decreased flow in red. Solid bars reflect the mean. Error bars in B are standard error. **p < 0.01. *p < 0.05.

https://doi.org/10.7554/eLife.05154.004
Figure 2—figure supplement 1
Experimental time-course.

We initially decided to analyze 1 min before and 1 min after handshake (analysis #1). We then noted, however, that this analysis may be inappropriate, because in the 60 s post handshake, an experimenter was still in the room for about 20 s with variance across subjects. This variance introduced potential imbalance across subjects. For example, a subject with a 14 s greet was then alone for an added 46 s of observation, yet a subject with a 29 s greet was then alone for an added 31 s of observation. To account for this potential imbalance, we added for each subject his/her individual greet time (20 ± 8 s) to each end (pre and post) of the observation. Thus, we end with ∼80 s before and after the handshake event. The results of these two analyses (120 and ∼160 s) were nearly identical (no effects lost or gained), with slightly stronger effects in analysis #1. We present analysis #2 in the manuscript.

https://doi.org/10.7554/eLife.05154.005
Figure 3 with 1 supplement
Humans sniff their own hands after handshake.

(A) Right and left hand changes in duration of face-touching following a greet. Duration change scores are after individual-baseline and condition-baseline normalization. The lettering under each pair of columns (e.g., F/F) reflects the ‘Subject gender/Experimenter gender’ interaction, respectively. The summation on the right is the interaction reflecting increased sampling of the right hand following within gender greets with handshakes, and increased sampling of the left hand following cross-gender greets without a handshake. (B) Three screen-shots depicting from left to right: a subject during baseline before the greet, then during handshake greet, and finally self-sampling after the experimenter leaves the room (see Video 3). (C) The spatial distribution of change in right-handed face-touching following the greet. (D) Latency to face-touch in the handshake (HS) and no-handshake (NHS) conditions. The figure contains only subjects who touched their face within the analysis time window. The 14 subjects with left hand continuously at face before during and after the greet were omitted from the figure. The dotted lines reflect the mean for each condition. Error bars are standard error. **p < 0.01.

https://doi.org/10.7554/eLife.05154.008
Figure 3—figure supplement 1
Reanalysis without correcting for condition-specific baseline.

The color legend reflects the type of interaction: subject gender (F/M), experimenter gender (F/M), and nature of greet with or without handshake (HS/NHS). The raw data for this analysis are in Supplementary file 1, and the ANOVA table is in Supplementary file 2. We conducted a repeated measures omnibus analysis of variance (ANOVA) with factors of subject gender (M/F), experimenter gender (M/F), and nature of greet (handshake/no-handshake), and a dependant repeated compact variable of exploration change time for right (shaking) and left (non-shaking) hands (hand). The ANOVA revealed no main effects or secondary interactions other than one highly significant interaction of subject gender × experimenter gender × greet × hand (F[1,145] = 12.75, p = 0.0005). This reflected a gender × experimenter gender × greet interaction for the right shaking hand (F[1,145] = 9.46, p = 0.0025), whereby both men and women equally (F[1,152] = 1.06, p = 0.3) increased apparent olfactory right hand exploration after shaking the hand of a same gender individual (change after handshake greet = 2.14 ± 8.1 s, change after no-handshake greet = −5.39 ± 15.3 s, t[71] = 2.69, p = 0.009), but not the hand of an opposite gender individual (a trend in the opposite direction, towards decreasing hand exploration: change after handshake greet = −4.85 ± 18.92 s, change after no-handshake greet = 0.94 ± 5.58 s, t[78] = 1.84, p = 0.07). In other words, individuals significantly increased right hand exploration only following the same gender greets that contained a handshake. In contrast, for the left non-shaking hand, a gender × experimenter gender × greet interaction (F[1,145] = 5.14, p = 0.02) reflected mostly trends, which moreover were in the opposite direction from the right hand. Specifically, both men and women equally (F[1,145] = 0.53, p = 0.47) did not change left non-shaking hand exploration after shaking the hand of a same gender individual (change after handshake greet = −2.85 ± 16.72 s, change after no-handshake greet = 1.64 ± 22.08 s, t[71] = 0.99, p = 0.33), yet had higher left non-shaking hand exploration after shaking the hand of an opposite-gender individual (change after handshake greet = 2.91 ± 11.84 s, change after no-handshake greet = −7.95 ± 28.33 s, t[78] = 2.26, p = 0.03). Note that this latter effect was not driven by increases in sampling after the handshake greet (+2.9 s), but rather by avoiding self-sampling after no-handshake greet (−7.95 s). Taken together, these data imply that after greeting individuals across gender without a handshake, humans may increase left non-shaking hand exploration, yet after shaking hands with individuals of the same gender humans robustly selectively increase investigation of only the hand that shook.

https://doi.org/10.7554/eLife.05154.009
Tainting with odors alters post-handshake hand sniffing.

(A) A schematic of the covert tainting device used for AND and EST. The non-shaking hand covertly activated a modified watch on the shaking hand that then emitted a plume of odor during handshake. (B) Face-touching behavior following tainting. Duration change scores are after individual-baseline normalization only. Error bars are standard error. **p < 0.01. *p < 0.05.

https://doi.org/10.7554/eLife.05154.012

Videos

Video 1
Humans often sniff their own hands.

This is an assortment of scored events from across the data (before, during, and after greet), demonstrating that humans often apparently sniff their own hands. Note that these are not the ‘best cases’, as typically subjects who engaged in very overt self-sampling did not later consent to use of their video in publication.

https://doi.org/10.7554/eLife.05154.006
Video 2
Hand sampling is concurrent with sniffing.

The video is from the control experiment that included a nasal cannula for nasal airflow recording. The airflow cursor is time-locked with the video. Note that when the subject was scored as sampling, he concurrently sniffed. A second example is in Video 4. This individual is obscured by pixalization to reflect requested level of privacy.

https://doi.org/10.7554/eLife.05154.007
Video 3
Humans sample the hand that shook.

Several greet events with ensuing behavior. The text in the upper left corner denotes the scored condition. Again, these are far from ‘best cases’, as typically subjects who engaged in very overt self-sampling did not later consent to use of their video in publication. Finally, some individuals are obscured by pixalization to reflect requested level of privacy.

https://doi.org/10.7554/eLife.05154.010
Video 4
Pronounced sniffing of the hand that shook.

An example from the control experiment that included a nasal cannula for nasal airflow recording. Although this may seem like a staged dramatization, it is not. This is raw data, with an explicit self-sample that occurred the moment the experimenter ended the greet and left the room. As the frozen image at the end highlights, this self-sampling behavior was perfectly timed with a pronounced sniff.

https://doi.org/10.7554/eLife.05154.011

Additional files

Supplementary file 1

Data and analysis. This excel file contains the analysis scheme, starting with raw duration values, and all the way through derivation of the data figures. Sheet 1 contains the initial data and its baseline correction. Sheet 2 contains the data separated by condition. Sheet 3 contains correction for condition. Sheet 4 contains the final data derivation. Sheet 5 contains data of the tainting experiment. Sheets 6 through 8 are a repetition of sheets 2 through 4 after deleting extremes from the condition-specific baseline.

https://doi.org/10.7554/eLife.05154.013
Supplementary file 2

ANOVA table for analysis without condition-specific correction. This excel file contains the ANOVA table for Figure 3—figure supplement 1. It reveals that the significant interaction is maintained when we do not correct for condition-specific baseline, and instead add it as an additional level in the analysis (with/without handshake).

https://doi.org/10.7554/eLife.05154.014

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)