Thermal imaging as an emerging technique to study proximate causes of behaviour: A review on current methods and future directions

Causal factors affecting behaviour fall into two categories. Ultimate causes, studied for example in behavioural ecology and evolutionary biology, involve the fitness of genotypes; they help explain why natural selection has favoured some behavioural strategies over others. Proximate causes, studied in the fields of cognition, motivation and emotion, involve individual-level perception, decision-making, value judgments and learning processes; they help explain why an individual is motivated to carry out particular behaviours at particular times. In furthering our understanding of proximate causes of behaviour, one way forward is the development of new methods to measure these internal processes of individuals, whether conscious or non-conscious. Subjective states cannot be measured directly, but substantial information can be gleaned about the processes of behavioural decision-making through measuring physiological processes linked to them. Ideally, such measures should be non-invasive and remote, in order not to affect the processes being measured. One of the technologies increasingly utilised to this end is thermal imaging, also known as infrared thermography. It is considered to hold substantial promise as a research tool for the future, including automated monitoring of animals in laboratories, zoos and veterinary clinics, on farms and in the wild. Thermal imaging of individual animals allows for remote measurement of the distribution of temperatures across a surface, such as an animal’s face or body, by an array of sensors detecting infrared radiation. These measurements are then converted to numerical temperature data, which often are visualised as heat map images (see Fig. 1 for an example). There is a wide range of different types of thermal cameras for different purposes and with different levels of image resolution. In most cameras used for scientific research, spatial resolution varies from 240 x 360 pixels to 768 x 1024 pixels, and temporal resolution varies from 8 to 200 images per second. The higher end of these allows for detailed measurements of free-moving animals at distances of a few metres to several tens of metres, depending on the size of the animal and the resolution required. Prices of thermal cameras have decreased substantially over
the years and are expected to continue to do so, which has added to the interest in developing methods to use them in a variety of settings. In the context of studying proximate causation of behaviour, the most common use of thermal imaging so far has been the study of the relationships between thermal physiology, ambient temperatures and behaviour (e.g. in laboratory rodents;
farm animals such as cattle, pigs and poultry, and wild animals including both endothermic and ectothermic species). Recent years have also seen a rapid increase in research to develop thermal imaging methods to measure physiological processes that are directly linked to emotional states, such as activation of the sympathetic nervous system during emotional arousal, which causes peripheral vasoconstriction and therefore a measurable reduction in peripheral temperature. Research by several groups including ours has demonstrated substantial potential to utilise such findings in the development of new methods to measure perception, emotion and motivation in a wide range of species. However, the increased availability of thermal cameras and the
aesthetically appealing nature of thermal images is a double-edged sword. While thermal imaging is a mature technology, its use in the development of new research methodology to measure physiological changes linked to causal processes is still at an early stage. One of the essential requirements for successful further development is an increase in interdisciplinary collaboration between researchers with backgrounds in ethology, physiology and cognition as well as thermal physics. Among the methodological research needs, one of the most important involves quantifying how other
physiological processes, such as digestion of recently eaten food, exercise and diurnal rhythm, affect surface temperature distribution in those parts of the face and body that are typically of interest in emotion studies. As an example, the ear pinna, in which peripheral vasoconstriction and the resulting temperature drop during emotional arousal have been found (e.g. in the dog and rabbit), also acts as part of the thermoregulatory vasoconstriction/vasodilation system in several species, leading to the prediction that there are species-specific threshold values for low and high ambient temperatures that would mask emotion-related effects in ear pinna temperature. Another major area in which methodological research is still needed concerns the effects of the environment can have on the temperature parameters of interest. Air currents, radiated heat from nearby animals, reflected heat from nearby surfaces as well as residual heat from recent touch by other animals or recent direct sunlight, are all examples of external factors that often have substantial confounding effects in temperature readings recorded by the thermal camera. Systematic, interdisciplinary studies of these effects is therefore one of the essential components of developing reliable thermal imaging methods for measuring motivation for behaviour.