Horses discriminate human body odors between fear and joy contexts in a habituation-discrimination protocol

Abstract

Animals are widely believed to sense human emotions through smell. Chemoreception is the most primitive and ubiquitous sense, and brain regions responsible for processing smells are among the oldest structures in mammalian evolution. Thus, chemosignals might be involved in interspecies communication. The communication of emotions is essential for social interactions, but very few studies have clearly shown that animals can sense human emotions through smell. We used a habituation-discrimination protocol to test whether horses can discriminate between human odors produced while feeling fear vs. joy. Horses were presented with sweat odors of humans who reported feeling fear or joy while watching a horror movie or a comedy, respectively. A first odor was presented twice in successive trials (habituation), and then, the same odor and a novel odor were presented simultaneously (discrimination). The two odors were from the same human in the fear or joy condition; the experimenter and the observer were blinded to the condition. Horses sniffed the novel odor longer than the repeated odor, indicating they discriminated between human odors produced in fear and joy contexts. Moreover, differences in habituation speed and asymmetric nostril use according to odor suggest differences in the emotional processing of the two odors.

Figure 1

Habituation–discrimination protocol. (a) Schematic representation of the experimental set-up. (b) Photographs showing sample presentation. Photograph courtesy of Plotine Jardat.

Figure 2

Example of a subject using her left or right nostril to sniff the sample. Photograph courtesy of Plotine Jardat.

Figure 3

Emotion ratings reported by the participants after watching the fear- and joy-inducing videos. (a) Ratings for ‘fearful’. (b) Ratings for ‘joyful’. Boxplots show the median, first and third quartiles. Permutation tests, ***p ≤ 0.001.

Figure 4

Habituation and discrimination to the samples. (a) Habituation and discrimination of all horses. (b) Habituation according to the type of odor (e.g., group). The graphs are extracted from the corresponding models presented in Table 2. The error bars represent the standard errors from the models. ***p ≤ 0.001, ns: p > 0.05.

Figure 5

Differential use of the left and right nostrils when sniffing odors. The graphs are extracted from the corresponding models presented in Table 2 (see “Methods” section). The error bars represent the standard errors from the models. *p ≤ 0.01,**p ≤ 0.001.