The incentive value learning theory states that the value assigned to a resource depends on the motivational state of an animal when it first encounters that resource. The present experiment investigated whether domestic hens, Gallus domesticus, assign different incentive values to a novel food depending on their deprivation state when they encounter this novel food. Hens were first trained to find a novel food type hidden under wood shavings in a large food dish. Then, after half of the animals were food-deprived and half left not food-deprived, all hens were re-exposed to the novel food in a small food dish in a second test arena. Next, all hens were food-deprived and then given access to the original food dish with wood shavings but no food in the original test arena. Hens that had been food-deprived before re-exposure spent more time looking, pecking or scratching at the wood shavings in the dish and pecked more at the dish than did the hens that were re-exposed while sated. The sated hens also tended to sit down longer in the test arena than did deprived hens. Thus, hens that had been deprived before re-exposure to the novel food searched more actively for the food than hens that had been sated before re-exposure, indicating a higher motivation. These results prove for the first time that incentive value learning occurs in domestic hens.

Early environmental experience can have significant effects on an animal's ability to adapt to challenges in later life. Prior experience of specific situations may facilitate the development of behavioural skills which can be applied in similar situations to later life. In addition, exposure to a more complex environment may enhance cognitive development (eg increased synaptic density), which can then speed the acquisition of new behavioural responses when faced with novel challenges (Grandin 1989).

It has been suggested that, during instrumental learning, animals are likely to react emotionally to the reinforcer. However, they may in addition react emotionally to their own achievements. Such reactions would be of interest in the context of the animal's capacity for self-awareness. We therefore devised a learning experiment to disentangle emotional reactions to a food reward and to own learning in young cattle.

In humans, excessive pessimism is strongly indicative of a depressive mental state. The negative outlook demonstrated by the depressed person is seen as the result of the depression in some way biasing the person's thinking in a negative direction (Beck 1967). This negative thinking is most clearly manifested as biases in attention to, and memory for, negative information (Gotlib & Krasnoperova 1998). Conversely, an individual in an elated, positive mood is likely to perceive and recall information in a more ‘optimistic’ manner and be less prone to focusing on negative stimuli. Here we have a situation where an emotional state (positive or negative affect) and a cognitive state (a bias for mood-congruent information) appear closely linked. Assuming a similar link in animals, we are trying to use operant training techniques to derive systematic tests of cognitive bias which may reveal subjective emotional states which are, themselves, very difficult to measure directly. The rationale behind the technique is to train animals to expect positive or negative outcomes to two training tones of differing frequency. Rats were trained on a go, no-go task where one tone signalled pending reward and the animals were required to make an operant lever press in order to receive it, and the other tone signalled a short, unavoidable period of white noise which the animals quickly stopped responding to. Thus, animals were trained to signal what they ‘expected’ to happen given a particular predictive cue. According to cognitive theories of depression, a depressed individual should show a greater level of expectancy of negative events, especially when presented with ambiguous stimuli, ie they should act pessimistically. Positive mood is associated with positive expectations, thus we would expect an individual in such a mood state to evaluate an ambiguous situation in a more optimistic way. We presented our trained animals with novel tones and recorded their responses before and after a period of housing in an ‘enriched’ environment which is often assumed to be a positive treatment for animals.

Development of the technique and preliminary results were presented.

Understanding the way in which animals perceive important features of their environment, and how this information may be stored and used, underpins the study of the impact of husbandry systems on animal welfare. Research into incentive contrast in laboratory animals has indicated that they can store and use information about stimuli associated with situations or resources. If cognitive processes are used, the relevant stimuli associated with a resource or event are encoded, allowing comparisons to be made when the key stimuli are changed. The presence and quality of this comparison, negative or positive, can be ascertained by assessing the change in the animal's level of appetitive or operant responding in order to gain access to the situation on the subsequent trial. We wished to determine whether the presence of a sudden change in appetitive response, such as speed of approach to an area following a change in incentive features of a resource or event in that area, is shown by the domestic chicken. This would indicate the presence of a cognitive comparison of incentives. The magnitude of such a change may also provide a measure of the welfare impact of the change. In the first experiment, 16 groups of four broiler chicks were trained to run a maze to an area where they received a high quality test diet. After training, the test diet of eight groups was changed to a low quality diet that all groups had been receiving in their home pens. Compared to the control unchanged groups, the experimental birds took increasingly longer to run the maze on successive post-change trials; the decrease reaching significance on the fourth post-change trial (P < 0.05). The behaviour of the chicks on the day of the change was indicative of frustration, with greater activity and redirected foraging shown (P < 0.05). Although there was no definitive evidence of cognitive processing which would have been shown by an immediate decrease in approach speed, this may have been due in part to the testing of groups of chicks. However, the occurrence of a frustration reaction indicated that some cognitive representation of food quality may have been present. In a second experiment, individual adult hens were trained to run a runway for access to food or water, while food- or water-deprived, respectively. Every fourth trial in a total of 16 trials, access to the resource was thwarted by a clear plastic lid on the dish. Time taken to traverse the runway was significantly longer on the two trials following this experience (P < 0.05), with no effect of repetition of thwarting (P > 0.05), indicating that the experience of thwarting was aversive. These results indicated that chickens appear to be able to store and use information on incentives. Additionally, with appropriate experimental design, an incentive contrast paradigm might be used to assess birds’ perception of the quality of changes in incentives.

The Concise Oxford Dictionary of Current English (Allen 1990) states that to compartmentalize is to ‘divide into components or categories’. In the context of the study of animal consciousness, compartmentalization may be considered to be an attempt to disregard certain beliefs in particular circumstances; to think about the same issue in different ways depending upon the context. For example, Fraser and Broom (1990) state that ‘the assessment of [animal] welfare can be carried out in a scientific way without the involvement of moral considerations … the process of deciding about morality and the process of assessing welfare are quite separate’. Yet is this proposed ‘switching on and off of beliefs in different contexts appropriate, let alone possible? Many animal scientists also use animals, and in doing so assume their animals are conscious. This assumption influences their interaction with them. It is not possible to ride a horse competently, for example, without assessing how it ‘feels’ about the signals one is transmitting; whether they are making the horse irritated, or whether it is calmly accepting them. Good riders, whether aware of it or not, are those who are sensitive to the horse's reactions to their movements. Recognized expressions of the horse's ‘feelings’ when ridden include rounding of the posture in the excited horse (Mills & Nankervis 1999) and shying in the frightened horse (Fraser 1992). In crediting animals with consciousness as an animal user, it is surely not logical to assume they are not conscious when behaving as a scientist. Thus compartmentalization may be an illogical attempt at parsimonious oversimplification, which is essentially contradictory to one's practical experience of animals, and therefore a hindrance in attempting to assess animal consciousness. Of course anthropomorphism must be avoided; it has limited value in the scientific study of animal behaviour and welfare. Furthermore, it could be considered demeaning to animals to assume that consciousness could only take the form that humans experience; that they are ‘humans in animal suits’. Yet it is equally wrong to ignore the logical objections to compartmentalization.

One of the basic components of animal perception and cognition is the understanding that there are invariant physical properties attached to objects in the world. The most commonly discussed aspect of object-concept is object permanence - the notion that objects are separate entities that continue to exist when out of sight of the observer (Piaget 1952; Krushinsky 1990). Pepperberg (1990) showed that different species of parrot are capable of reaching Stage 6 (Piaget 1952) but nothing is known about object permanence in young birds.

There is growing interest in exploring the cognitive abilities of animals, but the number of species studied is still very limited. Here, we would like to suggest that the domestic dog offers a very good example for investigating animal cognition.

Changes in plasma Cortisol concentrations in response to a stimulus have been used to indicate aversiveness. Hormone concentrations at any time point depend on the interaction between an animal's previous experience and its present physiological state. In order to assess how an animal's perception of, and thus its response to, a stimulus change as it gains experience of the stimulus, eight Welsh Mountain ewes were subject to 2h road transport daily on three occasions and then once again 4 days later. Differences in Cortisol concentrations in plasma samples taken every 15min were assessed using a general linear model and Tukey's test.

Recently I discussed a framework to study consciousness in animals (Van den Bos 2000).

This framework comprised:

• i) a hierarchical negative feedback model of the central nervous system in which behaviour is the control of input (perception) expressed as transitions of behavioural patterns (= motion) and as transitions of mental states (= consciousness; Van den Bos 1997). Consciousness is hypothesized to be a property of neuronal networks of self-organizing systems dedicated to dealing with rapidly-changing environments affording flexibility of behavioural patterning;

• ii) mental states - mental representations - as intentional states as described in folk psychology (feelings, desires; beliefs, insights, etc);

• iii) the separation of mental states into two components, an invariant part (‘the presence as such’; that) and a variant part (‘the way it looks’; how), which are separately related to the organization of the central nervous system, ie a ‘neuronal network’ and ‘momentary active connections within the neuronal network determined by input and output of this neuronal network’ respectively;

• iv) the idea that every mental state (psychological trait) - whether feelings, desires or beliefs - can be viewed conceptually as the product of a specialized neuronal network and information encoded in the network (neuronal input-output relations) which is expressed as a specific series of spontaneous behavioural patterns or in specific tests; and

• v) a phylogeny based on the invariant part, ie mental states, and the presence of a neuronal network.

It is my conviction that all animals, including invertebrates, possess a mind. This conviction is not based on scientific evidence. We are unable to observe the mind directly. According to the principle of parsimony, I assume that body and mind are two aspects of the same. This assumption predicts that it will always be possible to find a bodily explanation, implying that it will never be possible to perform an experiment that proves that a mind is present in any animal species, including human beings. As a result, it will never be possible to prove scientifically that my conviction that all animals possess a mind is correct or wrong.