Understanding Chicks’ Emotions: Are Eye Blinks & Facial Temperatures Reliable Indicators?

In commercial farming systems, chicks are reared without a mother. This absence of maternal influence can cause welfare problems when the chicks become older. Chicks imprint on their mothers they are young, and this mediates their stress and fear response. It is important to recognise problems early in the development of chicks to avoid welfare issues when they are older. One way to assess welfare is by measuring affective states. Research has shown chickens can display empathy, both towards their offspring and towards conspecifics. Measures of negative and positive affective states, either behavioural or physiological, could be good welfare indicators. This study employed non-invasive methods to measure affective states in laying hen chicks. Using video and thermal imaging, it analysed temperature changes in the peripheral areas and head region as well as changes in blinking behaviour before and after exposure to a stressor. The prediction was that the temperature would decrease in the eye and peripheral regions in response to a stressor and that the blinking rate would decrease. These changes would be indicative of a negative affective state. The results showed that the eye temperature as well as the blinking rate both decreased, whereas the temperature in the head region and the beak area increased. These results could be indicative of a negative affective state.


Manuscript contribution to the field 13
For objective evaluation, monitoring and management of poultry welfare, novel non-invasive 14 indicators, proxies, and iceberg indicators of stress is required. Using thermal imaging and video 15 imaging, we explored animal-based indicators such as eye blinks and facial temperatures of chicks as 16 measures of stress. Routinely collected video data from poultry commercial farms has the potential to 17 be digitized by Artificial Intelligence and the combination of indicators of stress be analyzed in the 18 development of welfare assessment protocols. Monitoring facial temperature variations via infrared 19 imaging and blinking frequency of chickens could be a sensitive metric that may offer insights on the 20 stressful environments. The identification of relevant poultry welfare indicators to improve welfare 21 assessment such as the iceberg indicators are essential to monitor welfare at both small scale and 22 large-scale industrial poultry production facilities. 23 Abstract 24 In commercial farming systems, chicks are reared without a mother. This absence of maternal 25 influence can cause welfare problems when the chicks become older. Chicks imprint on their mothers 26 they are young, and this mediates their stress and fear response. It is important to recognise problems 27 early in the development of chicks to avoid welfare issues when they are older. One way to assess 28 welfare is by measuring affective states. Research has shown chickens can display empathy, both 29 towards their offspring and towards conspecifics. Measures of negative and positive affective states, 30 either behavioural or physiological, could be good welfare indicators. This study employed non-31 invasive methods to measure affective states in laying hen chicks. Using video and thermal imaging, 32 it analysed temperature changes in the peripheral areas and head region as well as changes in 33 blinking behaviour before and after exposure to a stressor. The prediction was that the temperature 34 would decrease in the eye and peripheral regions in response to a stressor and that the blinking rate 35 would decrease. These changes would be indicative of a negative affective state. The results showed 36 that the eye temperature as well as the blinking rate both decreased, whereas the temperature in the 37 1 Introduction 40 In commercial farming systems, laying hen chicks are not brooded by a mother hen but in large 41 incubators in hatcheries and brooders on the farm (1). This means the chicks will never have a 42 mother figure and will not benefit from maternal care (2). Chicks are precocial animals, meaning they 43 are mobile from the moment they are hatched, but they still require maternal care in the first weeks of 44 their lives (3). This maternal care positively influences the behavioural development of the chicks, 45 for example, by imprinting certain behaviour related to vocalisations, feeding, and mediating fear and 46 stress responses (4). How the chicks are reared from hatching until laying age can influence the 47 welfare of the grown laying hens. Laying hen chicks are exposed to many stress factors in early life, 48 such as transportation, vaccination, manual sex sorting, not to mention the hatching in the incubators 49 itself (5). Earlier studies have found that mother hens can buffer the stress of their chicks and that 50 chicks show less fearful responses in the presence of their mother (2). A bird in a good state of 51 welfare should display a healthy body and a positive affective state and should be able to express 52 natural behaviour (6). Early negative experiences can have long-term effects on the development and 53 behaviour of the laying hens and can cause unwanted behaviour, like feather pecking (7). 54 However, natural brooding is not commercially feasible in domestic chicken farms (8). Brooding 55 hens have a lower feed conversion (9) and their growth rate is reduced (10). In addition, brooding 56 hens do not lay eggs, therefore, taking up space that could be used for hens that are producing eggs 57 (4). Thus, it is important to find other methods to fulfil the needs of the chicks that their mother 58 would provide for naturally and to improve the chicks' welfare. To develop a welfare platform with 59 the help of sensors and technology, many aspects can be researched, one of which is the assessment 60 of affective states. A deeper understanding of affective states in animals is needed to take the proper 61 measures required for positive states of welfare and to limit negative states welfare (11). 62 In animals, affective states can be measured through physiological components and behavioural 63 measurements, like skin temperature, head movements, and locomotion (12,13 studied the anxiety, the reaction to a perceived threat, and fear, the reaction to a known threat, of 76 broiler chickens in response to stocking density (19). It was found that high stocking densities 77 contributed to a reduction of the fear response and complex environment reduced anxiety. An 78 overview of affective states measured in avian animals with sensors can be found in Table 1. 79 Emotions are not individual components. In many social animals, the phenomenon of emotional 80 contagion has been found. Emotional contagion can be defined as the emotional state-matching of a 81 subject with an object (20). It means the individual is aware of the emotions of a conspecific animal 82 (11). In chickens, it was measured that the stress response of an observer chicken was greater when 83 the demonstrator chicken was handled roughly than when it was handled gently, suggesting the 84 presence of emotional contagion and empathy (21). Chickens do not have to experience a stressful 85 situation to feel stressed. The mere presence of a stressed conspecific can cause them to feel stressed 86 (11). 87 Important for emotional contagion is socially mediated arousal, which is the increased alertness to 88 respond that can occur when one animal perceives another's behaviour or physiology (22). 89 Behavioural indicators can be freezing responses and reduced maintenance behaviour, such as 90 preening. Physiological indicators can be heart rate variability, temperature variations, and hormonal 91 variations. A study about the presence of socially mediated arousal in mother hens, by Edgar et al. 92 (2015) (2), found that when chicks received a mild stress treatment, the mothers had an increased 93 heart rate, experienced stress-induced hypothermia, produced more vocalisations, increased standing 94 alert, and reduced preening behaviour. 95 Although this study does not focus on emotional contagion, but rather on specific affective states, it 96 is important to mention that chickens can feel for one another, which has implications for the welfare 97 of the chickens. 98 Blinking rate as a measure of emotions has been tested in several animal species. In horses, the 102 spontaneous blinking rate was measured as a possible indicator of stress. It was found that the 103 blinking rate first decreased and after a few minutes increased, a result similarly found in humans 104 (31). A study with crows showed that when crows are exposed to a threat, their blinking rate 105 This is a provisional file, not the final typeset article decreases, as they respond with a fixed gaze (32). Dogs showed an increased blinking rate in 106 response to fear and frustration, although this might have been a response to stress instead of these 107 specific emotions (33). 108 Blinking is necessary for clear vision, but it also temporarily impairs vision, as visual information 109 input is inhibited (34,35). When an animal blinks and closes its eye, it is essentially blind and can not 110 pay attention to its surroundings. A high blinking rate and long duration of blinks can cause 111 intermittent blindness, which leads to a reduced detection of threats (36). Chickens have mostly 112 monocular vision, with the field of view being around 300°, with just 30° overlap, thus allowing for 113 binocular vision. Gaze shifts are needed to obtain large changes in view, which are enabled by the 114 flexible neck and light head (37). A study tested whether songbirds were able to modify their 115 blinking behaviour when they were subjected to a potential threat. It was found that the birds 116 modified their blinking behaviour based on reactivity, and they could modify their blinking rate 117 based on a perceived risk (34). 118 Measured with the help of video cameras, the blinking behaviour could also be an interesting 119 indicator to explore affective states in chicks. When an animal is in a negative affective state, for 120 example, when it is frightened, its blinking behaviour might change. Due to the increased alertness, 121 the blinking rate might decrease. Blinking could therefore be a measure of affective state. 122

Thermal imaging 123
Another way to assess affective states in chickens is by measuring physiological responses, such as 124 temperature (38). Infrared thermography is used as a non-invasive technique to measure body surface 125 temperature which in turn can relate to stress and emotions in hens (28). Physiological components 126 can give indicators of emotional states that cannot be assessed verbally in animals. An animal under 127 stress will display cutaneous vasoconstriction, leading to a drop in skin temperature. This is paired 128 with an increase in core temperature, which in turn is followed by vasodilation, resulting in a post-129 stressor rise in peripheral temperature. This is known as emotional fever (28). 130 It has been shown that shank temperature of laying hens drops by 1-2°C for one to two minutes when 131 they are exposed to a frightening visual stimulus (39). 132 Infrared thermography, or thermal imaging, can detect infrared radiation emitted from an object (40). 133 It is used in many animal studies related to stress, emotional arousal, and animal welfare (15). To 134 measure skin temperature with infrared thermography, the skin needs to be bare. In chickens, the 135 wattles, comb, and head are already bare, which makes the chicken a good model for comparing 136 temperatures (25). 137 A study showed that when chickens were handled, stress caused an initial surface comb and eye 138 temperature drop of 2 °C and 0.8 °C respectively after one minute of handling, while core 139 temperature increased (41). Research in aggressive behaviour in pheasants found the head 140 temperature of both the attacker and the recipient of the attack first decreased before the attack, and 141 afterward showed an increase (42). Another study found that a rougher handling procedure also 142 caused a larger temperature shift, which suggests that skin temperature may be a way of quantifying 143 by measures of stressor intensity (25). 144

Hypothesis 145
In this study we examine whether a negative affective state can be observed in laying hen chicks in 146 response to a stressor by measuring physiological and behavioural components with the help of video 147 and thermal imaging. It is predicted the eye temperature would drop, whereas the overall head 148 temperature will rise as a sign of distress or fear. It is also predicted that the blinking behaviour will 149 change in response to a stressor, such as a reduced blinking rate to increase alertness. 150

2
Materials and methods 151 We examined the blinking behaviour and temperature changes in laying hen chicks as a measure of 152 affective states during a non-invasive experiment. The chicks were investigated for two weeks, 153 between November and December 2021, at the Research Facility CARUS of the Wageningen 154 University & Research. 155

Subjects 156
Fifty laying hen chicks of the breed Super Nick were used for this experiment. Upon arrival, they 157 were one day old. The experiment started when they were nine days old. The chicks were housed in 158 one pen of 2x4 meters (n=10) and two 4x4 meter pens (n=20). The chicks were only handled when 159 they were brought in, but not during the experiment. They received the necessary vaccinations upon 160 arrival. 161

Procedure 162
We assessed the chicks two or three times a week for two weeks. Their exact age on the days of 163 assessment was respectively: 9 days, 11 days, 15 days, 17 days, and 19 days. The experiment had 164 within-treatment control. The chicks were assessed before inducing a stressor and during or after 165 inducing a stressor. The assessment was done by taking thermal pictures and taking close-up videos 166 to measure the blinking behaviour. The stressors used were a playback of sounds and a visual 167 stressor. For the thermal pictures, only the visual stressor was used. The visual stressor was a pink 168 umbrella opening close to the pen. The umbrella was only opened and closed once per cage after 169 which thermal pictures were immediately taken. On days 9, 11, 15, 17, and 19 the chicks received the 170 umbrella stressor. During the assessment of the blinking behaviour, the chicks received two different 171 treatments. On days 9, 11, and 15, the chicks received only the visual stressor, the opening and 172 closing of the pink umbrella. On days 17 and 19, the chicks received only the sound stressor which 173 was a playback of the aggressive barking of a dog. This sound was played on a speaker (JBL Flip 2). 174 For this stressor, the chicks were filmed while the sound was playing. This sound was played for 10 175 minutes per cage during which the birds were filmed. It was measured whether these stressors evoked 176 a certain response, both physiological and behavioural. 177 For the thermal pictures, a thermographic camera was used, the FLIR 1020. The pre-stress thermal 178 pictures of all chicks were taken for 45 minutes. Fifteen pictures were taken in the cages where 20 179 birds were housed, and 10 pictures were taken in the cage where 10 birds were housed. Then stress 180 was induced. Immediately afterward, 40 thermal pictures of all the chicks were taken again for 45 181 minutes. Then stress was induced again. One hour after taking pictures after the stress treatment, 40 182 post-stress thermal pictures were taken. In total, 40 pictures were taken in each condition: pre-stress, 183 stress, and post-stress, 120 pictures per day. 184 In the experiment, we planned to measure the temperature of the eye, comb, head, and beak area, so 185 parts of the chicken are bare-skinned. However, the chicks did not have a comb yet, so we could only 186 measure the eye, beak, and head temperature. The thermal camera was handled manually. The head 187 This is a provisional file, not the final typeset article of the chicks was photographed, and the photos were manually analysed with the program FLIR 188 ResearchIR. In the program, the region of interest could be drawn manually and provided the average 189 and maximum temperature of the area (Figure 1).  stability. In addition, a GoPro Hero 9 camera with the frame rate of 240 fps was also deployed to 208 collect images of the chicks. This camera was used to capture more elaborate and in-depth details on 209 the face of the chicks. The blinking behaviour was measured before and after exposure to the 210 stressor. As the experiment was non-invasive, picking up the birds was not allowed, and, therefore, it 211 was not possible to assess the blinking behaviour of each chick from a particular position and 212 distance. Instead, around five birds in each pen were followed, and their eyes were filmed as closely 213 as possible for 1 to 3 minutes. The birds did not wear an identification tag, so the observer was not 214 able to identify each bird. However, the research ensured that the same bird was not filmed twice by 215 filming each chick in a different area of the cage. In addition, some birds could be identified by an  This study aimed to investigate whether a negative affective state could be measured by assessing 357 changes in blinking behaviour and temperature changes in the head region in response to stress 358 treatments. 359

Blinking behaviour 360
Chickens possess two eyelids and a nictitating membrane on each eye. A nictitating membrane is a 361 transparent third eyelid that is drawn horizontally across the eye. It can moisten the eye while 362 maintaining vision (43). It keeps the cornea clean and can provide mechanical protection for the eye 363 (44). Chickens do not typically blink with their upper and lower eyelid, but only with their nictitating 364 membrane. The term "full blink" is therefore not necessarily used to refer to a blink, but is used as a 365 term for a full closure of the eye. We looked at the eye closure as during this period they are fully 366 non-vigilant. When a bird blink using only its nictitating membrane, it essentially still has partial 367 vision. When a bird is stressed, thus needs full vigilance, it was expected blinking would still occur, 368 as it is important for cleaning the eye and maintaining vision. However, complete eye closure fully 369 impairs vision which is why it was decided to look at the frequency of these full blinks during stress 370 as well. 371 The obtained results demonstrate that chicks do not significantly blink less partial blinks when they 372 are stressed, but do display less full blinks when they are exposed to a stressor. This means the 373 alertness increases during stress treatment. During the stress treatments with the visual stressor, the 374 umbrella, the birds were respectively 9, 11, and 15 days old. In this period, the display of full blinks 375 was inhibited. The stress treatments with the dog barking noise occurred when the birds were 17 and 376 19 days old. During this treatment, both a slight reduction in partial blinks and a reduction in full 377 blinks were observed. The exposure of the chicks to the dog barking sound was an acute noise 378 treatment. A study on the effect of noise on chicks investigated the neuroendocrine interaction during 379 noise stress (45). Both the effects of acute noises and chronic noise were tested. Several 380 physiological changes were measured, as well as behavioural changes related to water and feed 381 intake. Loud noises of 80 and 100 dB caused more response during the acute noise treatment. For 382 chronic stress treatment, the noises of 60 dB also evoked a response on water and feed intake. The 383 present study did not measure the loudness of the dog barking sound used. It could be interesting in 384 future studies to measure whether the loudness of the noises played affects the blinking behaviour in 385 chicks. 386 Differences in the reaction to auditory and visual stress could be explained by the mechanisms of 387 how chickens process information and how they react to stress. Chickens have a lateralized brain, 388 which means information is processed differently in each brain hemisphere. The right hemisphere is 389 more involved with fear, stress, predator detection, global spatial attention, and small differences in 390 stimuli and the left hemisphere is more involved with the discrimination of objects from distracting 391 stimuli, detection of large differences between stimuli, and proximal spatial attention (46). In this 392 experiment, where the birds were exposed to auditory and visual stimuli that evoked a stress 393 This is a provisional file, not the final typeset article response, both hemispheres played an important role in processing information. Based on the number 394 of blinks, it could not be detected whether one stress treatment gave a larger stress response than the 395 other stress treatment, but the chicks did seem to react differently to the stressors in terms of full 396 body behaviour. Behavioural changes in the whole body of the chicks could be observed in response 397 to the stress treatments. These responses, such as increased wing flapping and increased vocalisations 398 during the Umbrella treatment and standing response and reduced preening behaviour during both the 399 Dog Barking treatment and the Umbrella treatment were not quantified during this study but could be 400 interesting to explore in further research. Other studies identify behavioural stress responses in chicks 401 as behavioural characteristics that differ from their normal performance. This could be aggressive 402 behaviour, such as feather pecking or changes in feeding activity (5). 403 In addition, Yorzinski (2016) and Beauchamp (2017) found research that many blinks occur during 404 gaze shifts, for example, when eating (35, 36). A study on the relationship between gaze shifts and 405 blinking rate in peacocks found that the animals timed their blinks during large gaze shifts to concur 406 with periods in which vision was already impaired. Blinking was inhibited mostly when they had to 407 perform many gaze shifts, increasing their alertness (35). A study where the blinking rate was 408 measured during feeding found that all blinks occurred when the birds made up and down 409 movements while feeding. The blinks also occurred during gaze shifts, when the birds moved their 410 heads from side to side to monitor their environment (36). The chickens blinked less often when they 411 were observing their environment than when feeding. During feeding, the blinks were shorter than 412 while monitoring. Despite the short blinks, the birds were still blind for nearly half the time when 413 feeding. 414 As blinks are highly correlated with head movements and gaze shifts, it was important to quantify the 415 number of blinks that did not occur during head movements. Out of the total number of blinks (PB 416 and FB combined), 39.9% occurred without a gaze shift or head movement. The results ( Figure 5) 417 show a higher number of blinks without gaze shifts during prestress than during both treatments. 418 They also show a higher amount of blinking without gaze shifts during the Dog Barking treatment 419 than during the Umbrella treatment. The results could indicate that blinking rate decreases in stressful 420 situations, independently from the head movements or gaze shifts. The duration of the blinks was not 421 measured. It could be interesting to assess whether the partial blinks were shorter during stress than 422 during prestress. 423 As the chicks received the same treatments on multiple days (Umbrella treatment on days 9, 11, and 424 15 and Dog Barking treatment on days 17 and 19), there is a possibility that habituation may have 425 taken place for these stress treatments. The results from the statistical analysis showed there was a 426 significant difference in the number of blinks (Partial Blinks and Full Blinks combined) between 427 days 9 and 19 (p=0.024), between days 11 and 17 (p=0.011), and between days 11 and 19 (p=0.002). 428 However, the chicks were given different treatments on these days, which means effects cannot be 429 attributed to habituation, but the different stress treatments. No significant differences were found in 430 the number of blinks between the days on which the same stress treatment occurred. The chicks 431 would have to be assessed for a longer period to observe whether habituation occurs. 432

Thermal imaging 433
This study assessed the changes in temperature in the head regions in response to a stressor. Chickens 434 are homeotherm animals, meaning they keep a high and constant body temperature (47). However, 435 when an animal experiences a certain affective state, such as fear or happiness, the body displays a 436 physiological change (48).
The eye temperature showed a slight reduction in response to a stressor, though not significantly. 438 This reduction in eye temperature as a response to a stressor was found in other studies where 439 emotional contagion was measured within domestic chick broods (22), and in mother hens and chicks 440 that were exposed to a stressor (24). A reduced eye temperature is an indicator of stress-induced 441 hyperthermia (SIH) (22). Stress-induced hyperthermia, or emotional fever, is an acute stress response 442 mediated by the activation of the hypothalamic-pituitary-adrenal axis (HPA-axis) in combination 443 with sympathetic-adrenal medullary (SAM). This causes a release of stress hormones, such as 444 cortisol, corticosterone, and temperature changes (49, 50). 445 In addition to the changes in eye temperature, the average beak temperature also changed in response 446 to the stress treatments. The beak of birds is made of keratin. It is a very versatile organ, playing a 447 role in many functions, such as feeding, drinking, preening, manipulating, nesting, and fighting (51). 448 It possesses many nerve endings, is highly vascularised, and plays a role in thermoregulation (51). In 449 toucans, the beak acts as a highly efficient thermoregulatory device and can exchange heat up to 450 400% of the resting heat production (52). Therefore, changes in beak temperature can be an 451 interesting way of measuring affective states. 452 Feathers of birds are good insulators, trapping heat between the skin and the feathers. Although this 453 is a beneficial trait for birds in colder climates, it also makes losing heat in warmer climates more 454 difficult and makes the animals more susceptible to heat stress (53). A rise in temperature in the core 455 body, for example, in response to a stressor, means this heat also must be dissipated to the outer 456 environment. When a bird needs to get rid of heat, the beak and legs play a very important role (54). 457 Few studies have looked at the beak region temperature as a measure of stress. The present study 458 found that the beak temperature significantly increased during the post-stress period, one hour after 459 stress was induced. It is possible that due to the rise in core temperature, more heat had to be 460 dissipated, which was done through the beak. Sometimes, the chicks had just drunk water before the 461 thermographic pictures were taken. The results indicated a lower temperature in the beak, and, thus, a 462 reduced mean temperature. 463 The head temperature of the chicks remained constant during the three levels of stress. The feathers 464 on the head region of the chicks may have had an influence. The skin needs to be bare, which is not 465 the case with the chicks' heads. 466 The age of the birds has a significant influence on the eye temperature, the beak temperature, and the 467 head temperature. Although the thermal camera was calibrated every day of use, there is a possibility 468 environmental temperature played a role in the reduction of temperature in these three areas. As the 469 birds got older, the environmental temperature reduced. At the start of the measurements, when the 470 birds were nine days old, the environmental temperature had a maximum of 31.2°C. At the end of the 471 measurement period, when the birds were 19 days old, the maximum environmental temperature was 472 27.9 ° C . However, it is also possible the birds became less stressed, causing a smaller increase in core 473 temperature, as they got older and adapted to the treatment. 474 In most studies involving thermographic imaging, the animals are photographed at a set distance, for 475 example, one meter (55,29,56). As part of the study design, the chicks were always able to walk 476 freely inside the cages and were not restrained. Thus, the thermographic pictures were taken at 477 various distances and locations for each bird. This may have influenced the temperature of the chicks. 478 This is a provisional file, not the final typeset article To validate the affective states measured in the laying hen chicks, studies on cortisol in feather and-or 479 blood-based measures are needed. Acute and chronic stress can be assessed by measuring cortisol 480 and corticosterone in, for example, feathers or blood (57,58). The advantage of blood measures is that 481 hormonal changes in blood can happen quickly in response to a stressor, making it an effective way 482 of measuring negative affective states (59). In feathers, corticosterone can also be measured but the 483 levels do not change instantaneously, making it more difficult to assess affective states (59). 484

485
During this study, it was investigated whether affective states namely fear as an emotion could be 486 assessed in laying hen chicks with the help of video and thermal imaging. The results showed that 487 chicks display more full blinks before a stressor and that their blinking without gaze shifts decreased 488 during a stressor. The reduction in the full blinking rate could indicate a negative affective state. The 489 beak temperature changed significantly during the three different stress levels, in particular during 490 post-stress. Stress-induced temperature changes in the beak and eye region of the chicks were 491 significantly correlated to increase as the birds grew older. Future studies are warranted to explore 492 whether the environmental temperature could play a role in the variation of temperatures or whether 493 the chicks adapted to the stressors, causing less fearful responses. This study presented a quantitative 494 examination of both physiological and behavioural changes in response to a stressor, which could be 495 indicative of a negative affective state. For future research, feather or blood-based cortisol and 496 corticosterone measurements are needed for validation of the negative affective states. With the 497 advancements in AI enabled technology and sensors, blinking behaviour and the facial surface 498 temperature can serve as new indicators for poultry welfare. showing the correlation between the age of laying hen chicks to the total blinks and the stress type 506 applied on the laying hen chicks to the head movement without blinks are available as figures S1 and 507 S2 respectively. The results showing the correlation between the laying hen chick's head average 508 temperature to stress level and age of the chicks can be found as Supplementary graphs Figure S3  509 and S4 respectively. 510

7
Conflict of Interest 511 The authors declare that the research was conducted in the absence of any commercial or financial 512 relationships that could be construed as a potential conflict of interest. Funding 518 The authors thank the Next Level Animal Sciences program of the Wageningen University & 519 Research for partial financial support for this project. 520 521