Facial blushing and feather fluffing are indicators of emotions in domestic fowl (Gallus gallus domesticus)

Abstract

The study of facial expressions in mammals provided great advances in the identification of their emotions and then in the comprehension of their sentience. So far, this area of research has excluded birds. With a naturalist approach, we analysed facial blushing and feather displays in domestic fowl. Hens were filmed in situations contrasting in emotional valence and arousal level: situations known to indicate calm states (positive valence / low arousal), have rewarding effects (positive valence / high arousal) or induce fear-related behaviour (negative valence / high arousal). Head feather position as well as skin redness of comb, wattles, ear lobes and cheeks varied across these situations. Skin of all four areas was less red in situations with low arousal compared to situations with higher arousal. Furthermore, skin redness of the cheeks and ear lobes also varied depending on the valence of the situation: redness was higher in situations with negative valence compared to situations with positive valence. Feather position also varied with the situations. Feather fluffing was mostly observed in positively valenced situations, except when hens were eating. We conclude that hens have facial displays that reveal their emotions and that blushing is not exclusive to humans. This opens a promising way to explore the emotional lives of birds, which is a critical step when trying to improve poultry welfare.

Fig 1. Profiles of a hen showing different feather positions and skin redness, and the four regions of interest where redness was sampled.

(a) Illustration of the four region of interest (arrows). (b) Locations where skin redness was sampled (black squares, 10 x 10 pixels squares). (b,c) Location where feathers’ position was observed (blue rectangle). Figures illustrate Eden (Meusienne hen) with (a, b) feathers labelled as sleeked or (c) fluffed and with (a) low or (b, c) high skin redness. Pictures are extracted from (a) Resting, (b) Capture, and (c) Dustbathing situations.

Fig 2. Dendrogram (left hand side of the figures) and table from hierarchical cluster analyses of the redness of the skin for the four ROIs.

(a) M-hens. (b) P-hens. The heatmap contains the tested situations (Alert, Capture, Reward… in red, blue, green…) in rows and ROI (wattles, comb, cheeks, ear lobes) in columns with red colour indicating high-level of redness, blue colour low-level, and yellow colour being between the two. Tables indicate for each situation the number of hens in each cluster (M-hens, n = 7, P-hens, n = 10).

Fig 3. Median and interquartile range of the redness of the skin for wattles, comb, cheeks and ear lobes depending on the situation.

(a) M-hens. (b) P-hens. Situations with different subscripts differed significantly (GLMM on log2 transformed data with Tukey adjustment for two-by-two comparisons).

Fig 4. Redness (median and interquartile range) of the skin for wattles, comb, cheeks and ear lobes depending on the arousal of the situation.

(a) M-hens. (b) P-hens. Low arousal includes Feeding, Resting and Preening; High arousal includes Capture, Alert, Dustbathing and Reward. ****: p<0.001 (GLMM on log2 transformed data with Tukey adjustment for two-by-two comparisons).

Fig 5. Redness (median and interquartile range) of the skin for wattles, comb, cheek and ear lobes depending on the valence and arousal of the situation.

(a) M-hens. (b) P-hens. Negative valence / high arousal (V-/A+) includes Capture and Alert, positive valence / high arousal (V+/A+) includes Dustbathing and Reward, positive valence / low arousal (V+/A) include Feeding, Resting and Preening). Situations with different subscripts differed significantly (GLMM on log2 transformed data with Tukey adjustment for two-by-two comparisons).