Horses Cross-modally Recognize Women And Men Part 2

The videos were validated by 20 persons who categorized women and men with 100% accuracy. The vocal parameters of the recordings were analyzed using PRAAT v.5.0.3 (http://www.fon.hum.uva.nl/praat/). Fundamental frequencies (F0) were calculated using the PRAAT algorithm Pitch. 

Accuracy and memory are two very important aspects of human cognitive abilities. Accuracy refers to the correctness or precision in a certain field, while memory refers to a human being's ability to remember some things, characters, and events in long-term life and learning. There is an inseparable relationship between the two.

First, accuracy is critical to the identification and preservation of memory. Only accurate information can allow people to better understand and remember. If there are errors and biases in the information, it is easy for people to get confused and lose the meaning of their memory. For example, when learning historical knowledge, you need to understand the real historical facts. If there are biases and errors, it will lead to students' misunderstandings and memory errors.

Secondly, memory also helps improve accuracy. By constantly memorizing and retrieving knowledge, we can continuously deepen our understanding and knowledge of this field. At the same time, memory also helps to capture details and check for omissions. Through continuous testing and comparison with accuracy, people's knowledge can be made more accurate and complete.

Therefore, accuracy and memory are mutually reinforcing and inseparable. Only the two-pronged approach of accuracy and memory can give people a greater sense of accomplishment and security in learning and life. Therefore, when facing challenges in life and study, we should strive to improve our accuracy and memory so that we can better adapt to the environment and learn and develop. It can be seen that we need to improve memory, and Cistanche deserticola can significantly improve memory, because Cistanche deserticola can also regulate the balance of neurotransmitters, such as increasing the levels of acetylcholine and growth factors. These substances are very important for memory and learning. In addition, Meat can also improve blood flow and promote oxygen delivery, which can ensure that the brain receives sufficient nutrients and energy, thereby improving brain vitality and endurance.

Click Know to improve short-term memory

The mean F0 for the men's voices was 140 Hz (+SE=24), while the mean F0 for the women's voices was 215 Hz (+SE=19). The voice was played from the loudspeaker at a level of 60-70 dB according to the location of the horse.

Data collection.

Gaze responses. The horses' gaze direction towards the two different projection screens during playback events was analyzed. Horses were considered to be looking at a video if they turned their head towards the projection screen. One camera was placed under each projection screen. Horses were considered to look at the video when their faces were directed towards the camera such that the observer could see their opposite eye. Thus, in this preferential-looking paradigm, the gazes towards the congruent and incongruent videos for each trial were analyzed.

To avoid bias due to lateralization, horses were excluded from the analysis if they did not meet a criterion. Indeed, some horses looked almost exclusively at one projection screen (either placed on their left or their right, independent of the congruence of the video presented). To avoid this bias, the following criterion was established: during the test sessions, if a horse directed less than 10% of the total gaze duration to one side, they were excluded from the analysis. Therefore, four additional horses were excluded from the analysis.

Next, we analyzed three variables to investigate the gaze responses and calculate the preference indexes: (INC−CON)/(INC+CON).

1. Gaze duration: the abbreviation INC corresponds to the total gaze duration directed toward the incongruent videos, and the abbreviation CON corresponds to the total gaze duration directed toward the congruent videos.

2. Latency to the first look: the abbreviation INC corresponds to the duration from the beginning of the trial to the first look directed toward the incongruent video, and CON corresponds to the duration from the beginning of the trial to the first look directed toward the congruent video.

3. Number of looks: the abbreviation INC corresponds to the number of looks directed towards the incongruent videos and the abbreviation CON corresponds to the number of looks directed towards the congruent videos.

For each individual, the mean of the indexes obtained for each trial was calculated to obtain a mean preference index. The mean preference indexes were between 1 and 1. If the mean preference index equaled 0, the horse looked equally at the congruent and incongruent videos.

All the videos were analyzed by the same observer. The observer was blinded to the voice played and analyzed the videos without sound. A second observer analyzed 20% of the videos. The interobserver reliability was calculated with the interclass correlation coefficient (ICC)34. The ICC for the gaze duration had a lower bound of 0.91 and an estimate of 0.94, which indicated excellent interobserver reliability. The ICC for the latency to the first look had a lower bound of 0.99 and an estimate of 0.99, which is also considered excellent interobserver reliability. The ICC for the number of looks had a lower bound of 0.85 and an estimate of 0.90, which is considered good interobserver reliability.

Additional behaviors. In addition, the following behaviors were also recorded to assess the emotional response of horses: defecating, shaking their head, pawing the ground, and rearing. All horse vocalizations were also recorded. However, too few horses engaged in these additional behaviors to permit statistical analysis (defecating: N=3, shaking their head: N=2, pawing the ground: N=1, rearing: N=0, vocalizing: N=3).

Physiological responses. For each trial, the differences in the mean heart rate between the last 5 s and the first 5 s were calculated. Calculating this difference allowed us to determine whether the heart rate of horses increased or decreased while hearing the different voices. Heart rate data for two individuals were missing due to technical issues with the monitoring system.

Statistical analysis.

The statistical analysis was carried out using RStudio 3.3.6 statistics software. A comparison was made between the mean preference indexes and the theoretical value corresponding to a random choice obtained by chance (mu=0). The normality of the distribution of variables was assessed graphically using the function qqPlot in the package car. Data were analyzed with a parametric test: a bilateral sample Student's t-test using the function t.test. The heart rate responses to the voices (man's or woman's voice) were analyzed with a paired Student's t-test. The significance threshold throughout this study was set at P<0.05.

Ethical note.

We had permission to use animals and humans for experimental purposes as the study was approved by the Val de Loire Ethical Committee (Authorization number: CE19-2021-3011-1, CEEA VdL, Nouzilly, France). Animal care and experimental treatments complied with the French and European guidelines for the housing and care of animals used for scientific purposes (European Union Directive 2010/63/EU) and were performed under the authorization and supervision of an official veterinarian of the Département d'Indre et Loire (France). The subjects were not food-deprived during the experiment. They lived in social groups and had daily access to an outside paddock. The experiment collected only noninvasive data. All methods were performed according to relevant guidelines and regulations. Written informed consent was obtained from all the volunteers for their study participation and written informed consent for publication of images in an open-access was obtained for two of the participants.

Results

The mean preference index per individual was (INC − CON)/(INC+CON) for each trial. On the plot: for the gaze duration (Fig. 3a) and the number of looks (Fig. 3c), the closer the mean preference index is to −1, the longer horses looked at the congruent video, thus negative values indicate a preference for the congruent stimulus; for the latency of the first look (Fig. 3b), the closer the mean preference index is to 1, the faster horses looked at the congruent video, thus positive values indicate a preference for the congruent stimulus.

The mean preference indexes for the gaze duration were significantly different from 0 (Student's test: t= −2.212, P=0.036; Fig. 3a). Horses directed their gaze more towards the congruent video than the incongruent video. The means indexes for the latency to the first look were also significantly different from 0 (Student's test: t=-2.522, P=0.018; Fig. 3b). Horses directed their first looks more towards the congruent video than the incongruent video. The means indexes for the number of looks were significantly different from 0 (Student's test: t= −2.4254, P=0.023; Fig. 3c). Horses looked more frequently towards the congruent video than the incongruent video.

The change in the heart rate of horses did not significantly (NS) differ between women's and men's voices (Student's test: t= −0.079, P=0.869; Fig. 4).

Discussion

Horses gazed longer towards the congruent video: while hearing a woman's voice, horses gazed more toward the woman's video, and while hearing a man's voice, they gazed more toward the man's video. Moreover, horses were more likely to look first toward the congruent video, and they looked at the congruent video more frequently. 

Therefore, horses seem to associate women's voices with women's faces and men's voices with men's faces, suggesting that they utilize cross-modal cues to recognize humans. The heart rates of horses did not differ between the two types of voices.

In this experiment, horses looked preferentially at the congruent stimuli. In similar preferential-looking paradigms, some studies have indicated that subjects looked more toward the congruent stimuli, while others found the opposite. Our results are in line with a study by Proops and McComb done in 201214, which investigated the cross-modal categorization of familiar humans; in this study, horses spent more time looking at congruent stimuli. 

Conversely, in the study investigating the cross-modal categorization of children and adults15 and the one investigating the cross-modal categorization of human emotions12, horses spent more time looking at the video that was incongruent with the sound. A possible explanation for this discrepancy could be related to the nature of the stimuli, especially to the emotion that they induced. In our experiment, the two types of stimuli did not seem to induce any specific emotion, based on the heart rate response. 

In contrast, in Trösch and collaborators' study12, the heart rate of horses increased when they heard angry vocalizations, suggesting that these stimuli may have worried or surprised horses. In Jardat and collaborators' study15, the heart rate of horses increased when they heard children's voices, as the horses used in this study had never seen a child in their whole life, they might have been more stressed or surprised by voices with unfamiliar characteristics. This explanation is supported by another study in dogs33 that investigated their ability to cross-modally categorize humans according to sex. 

On the one hand, dogs living with both a woman and a man (and thus familiar with both sexes) looked more at the congruent person; on the other hand, when living with only one person, either a woman or a man, they looked more at the incongruent person. 

Thus, subjects might preferentially look at the congruent video when the stimulus does not induce any specific emotion (e.g., when stimuli are familiar) and look at the incongruent video when one stimulus induces worry or surprise. However, this potential explanation needs further investigation.

Furthermore, in our study, we tried to avoid inducing biases. First, we used various faces and voices; thus, the results are not individually dependent. Second, this study was a playback experiment, and the voices played while projecting the videos were not the original voices of any of the two persons in the videos presented to avoid the possibility that horses might match the mouth movements to the voices. However, additional investigations are needed to better understand which cues horses use to categorize woman and man stimuli. 

Women's and men's faces generally differ on numerous characteristics, such as shape or skin texture35, and their voices differ not only in frequency but also phonetically36. Horses could potentially form a holistic representation that takes all of these characteristics into account to categorize women and men. In a previous experiment, we showed that horses recognize a familiar human face in photographs despite modifications such as changing the color to black and white, changing the angle, hiding the eyes, or using different hairstyles11, which suggests that horses employ a holistic process for facial recognition. Nevertheless, the categorization of individuals into women and men may be based on a few specific cues and this study has limitations. We presented 6 women's faces and 6 men's faces, twelve people to have different people in each trial. 

However, the participants were all Caucasian in the same age range, and, as an example, all women had long hair, unlike men. Horses could simply associate high vocal frequencies with longer hair to distinguish between the women and men presented in this experiment. It would be interesting to generalize the result with other people.

Nevertheless, our study has some limitations. We chose to calculate indexes (INC−CON)/(INC+CON) to take into account the variability of the total gazing time between horses. Then, we compared these indexes to a theoretical value (zero), as had been done in previous studies using the same protocol12,15. This approach prevents us from addressing potential confounding variables such as the side of the person who stood next to the horses and the order of presentation of the stimuli. These variables were counterbalanced beforehand between horses to avoid any bias, however, to not include these factors in the statistical analyses is a limitation of our approach and we must remain careful about the interpretation of the results.

Conclusion

Horses can cross-modally recognize women and men: they associate women's faces with women's voices and men's faces with men's voices. Future investigations are needed to determine which characteristics horses use to categorize humans and whether they use simple cues or a more holistic process. These findings provide a novel perspective that may allow us to better understand how horses perceive humans. 

Indeed, they reveal that horses can effectively cross-modally categorize women and men. Specifically, that suggests that horses may generalize their experiences with one person to other people in the same category (i.e. women or men). Horses remember previous interactions with humans and act consequently in future interactions37. 

In this study, the use of a reward during a learning task induced positive reactions toward the human, and increased contact and interest toward that human several months later. It would be interesting to explore how horses use this information to adjust their behavior towards different individuals depending on their category membership. This knowledge could enhance horse training techniques and enhance the safety and welfare of both horses and humans.

References

1. Ringhofer, M. & Yamamoto, S. Domestic horses send signals to humans when they face an unsolvable task. Anim. Cogn. 20(3), 397–405 (2017). 

2. Ringhofer, M., Trösch, M., Lansade, L. & Yamamoto, S. Horses with sustained attention follow the pointing of a human who knows where food is hidden. Sci. Rep. 11(1), 1–9 (2021). 

3. Sabiniewicz, A., Tarnowska, K., Swiatek, R., Sorokowski, P. & Laska, M. Olfactory-based interspecific recognition of human emotions: horses (Equus ferus caballus) can recognize fear and happiness body odor from humans (Homo sapiens). Appl. Anim. Behav. Sci. 230, 105072 (2020). 

4. Schrimpf, A., Single, M. S. & Nawroth, C. Social referencing in the domestic horse. Animals 10, 164 (2020). 

5. Takimoto, A., Hori, Y. & Fujita, K. Horses (Equus caballus) adaptively change the modality of their begging behavior as a function of human attentional states. Psychologia 59, 100–111 (2016). 

6. Trösch, M. et al. Horses prefer to solicit a person who previously observed a food-hiding process to access this food: a possible indication of attentional state attribution. Behav. Process. 166, 103906 (2019). 

7. Trösch, M., Bertin, E., Calandreau, L., Nowak, R. & Lansade, L. Unwilling or willing but unable: can horses interpret human actions as goal-directed? Anim. Cogn. 23, 1035–1040 (2020). 

8. Trösch, M. et al. Horses feel emotions when they watch positive and negative horse–human interactions in a video and transpose what they saw to real life. Anim. Cogn. 23(4), 643–653 (2020). 

9. Jardat, P. & Lansade, L. Cognition and the human-animal relationship: A review of the sociocognitive skills of domestic mammals toward humans. Anim. Cogn. 25(2), 369–384 (2021). 

10. Lansade, L. et al. Female horses spontaneously identify a photograph of their keeper, last seen six months previously. Sci. Rep. 10, 6302 (2020). 11. Lansade, L. et al. Human face recognition in horses: data in favor of a holistic process. Front. Psychol. 

11, 575808 (2020). 

12. Trösch, M. et al. Horses categorize human emotions cross-modally based on facial expressions and non-verbal vocalizations. Animals 9, 862 (2019).

For more information:1950477648nn@gmail.com