A cross-species judgement bias task: integrating active trial initiation into a spatial Go/No-go task

Judgement bias tasks are promising tools to assess emotional valence in animals, however current designs are often time-consuming and lack aspects of validity. This study aimed to establish an improved design that addresses these issues and can be used across species. Horses, rats, and mice were trained on a spatial Go/No-go task where animals could initiate each trial. The location of an open goal-box, at either end of a row of five goal-boxes, signalled either reward (positive trial) or non-reward (negative trial). Animals first learned to approach the goal-box in positive trials (Go) and to re-initiate/not approach in negative trials (No-go). Animals were then tested for responses to ambiguous trials where goal-boxes at intermediate locations were opened. The Go:No-go response ratio was used as a measure of judgement bias. Most animals quickly learned the Go/No-go discrimination and performed trials at a high rate compared to previous studies. Subjects of all species reliably discriminated between reference cues and ambiguous cues, demonstrating a monotonic graded response across the different cue locations, with no evidence of learning about the outcome of ambiguous trials. This novel test protocol is an important step towards a practical task for comparative studies on judgement biases in animals.

(Supplementary Fig S1 for Batch 2 set-up). The experimenter operated the opening of the goal-boxes and the movement of the trial initiator from behind the wall with the goal-boxes.
The goal-boxes served as spatial cues, but the blue and yellow pattern on the inner side of the lid (same pattern in each goal-box) that was only visible when the goal-box was open, and the 'clack' sound when opening the lid served as additional visual and auditory cues, respectively. Each goal-box could only be opened and closed manually by the experimenter rotating a metal stick connected to The trial initiator was a green plastic 2 l bottle filled with approximately 0.5 l of water suspended from the ceiling at a height of 1.45 m, which could be raised and lowered by the experimenter. If the animal touched the bottle with its nose/head, the bottle was raised out of reach for two seconds. This raising of the bottle indicated a correct trial initiation and allowed the horse to focus on the goal-box cues before making a decision either to go or to re-initiate. each goal-box, a perforated floor concealed inaccessible rewards to prevent odour cues from indicating the presence or absence of reward. A small chute at the back allowed for delivery of reward into the goal-box (45 mg chocolate flavoured Dustless Precision Pellets®, Bio-Serv, New Jersey, U.S.A.).
The trial initiator was a nosepoke operandum (H21-09R, Coulbourn Instruments, Massachusetts, U.S.A.), connected to the Habitest Modular System® (Coulbourn Instruments), with an opening of 2.5 cm in diameter. The nosepoke contained an infrared (IR) beam and pairs of red, green, and yellow LEDs, of which only the green and yellow were used. The nosepoke was programmed so that by default both the yellow and green sets of LEDs were on, signalling that the initiator was available. If the animal broke the IR beam with its nose, all LEDs switched off to signal a correct initiation and that the initiator was briefly deactivated. One second after the animal had removed its nose from the nosepoke, the green LEDs switched on to signal that the initiator was available again to initiate a new trial, while the yellow LEDs instead flashed at 4 Hertz for a further 4 s before remaining on to signal to both the animal and the experimenter the maximum trial length (5 s). If an animal remained with its nose in the nosepoke, or repeatedly nose-poked before the 1 s time-out was up, the initiator stayed inactive until the animal had removed its nose for 1s. The 1 s time-out should allow the animals to focus on the goal-box cues before making a decision either to go or to re-initiate. Supplementary Fig S3. Goal-holes and test apparatus of the mice. The row of holes or goal-boxes (a) that served as the spatial cue when training and testing mice. Only five of the holes were used in the current study: every second hole starting from the outmost hole. Overview of the test arena (b) with the goal-holes at the top and the nose poke apparatus used as trial initiator at the bottom.

Mice
Mice were tested and trained under red light in three virtually identical apparatuses situated in a test room adjacent to the housing rooms. Each apparatus consisted of a rectangular test arena (l x b x h: 40 x 25 x 25 cm) with an additional V-shaped area at one end (depth 10 cm) containing the trial initiator and a row of nine identical goal-holes ( Supplementary Fig S3), resulting in a distance of 50 cm between the initiator and the central goal-box. Of these nine goal-holes (⌀ 1.5 cm, 3.7 cm apart from each other and beginning 2.5 cm above the apparatus floor) only five were used for the purpose of the current experiment. Each goal-hole was blocked by a sliding door which the experimenter could slide upwards to reveal the hole. A barrier above each hole ensured that doors could be opened quickly but would remain in their guides. The experimenter operated the apparatus from behind the goal-boxes. Water was available ad libitum via two water bottles attached at the sides of the arena.
The goal-holes (hereafter goal-boxes) served as the spatial cue in the task, but a white LED (connected to a 9v power source) placed behind each hole and the 'clack' sound when the door hit the barrier served as additional visual and auditory cues, respectively. Each goal-box could only be opened by the The operation of the initiator differed slightly between batches due to temporary technical problems during Batch 1.
 Batch 1: Only the green LEDs were used. The nosepoke was programmed so that by default the green set of LEDs were on signalling that the initiator was available. If the animal broke the IR beam within the nosepoke, all LEDs switched off to signal a correct initiation and that the initiator was briefly deactivated. After 1 s, the green LEDs switched on again to signal that the initiator was available again so the animal could opt for a new trial.
 Batch 2: The green and yellow LEDs were used. The nosepoke was programmed so that by default the both green and yellow sets of LEDs were on signalling that the initiator was available. If the animal broke the IR beam within the nosepoke, all LEDs switched off to signal a correct initiation and that the initiator was briefly deactivated. One second after the IR beam became intact again, i.e. the animal had removed its nose from the nosepoke, the green LEDs switched on to signal that the initiator was available again so the animal could opt for a new trial. The yellow LEDs instead flashed on and off every 250 ms for a further 4 s before remaining on to signal to both the animal and the experimenter the maximum trial length (5s). Any re-break of the IR beam at any point in this process reset the process from the beginning. In this way, if an animal remained with its nose in the nosepoke, or repeatedly nose-poked before the 1 s time-out was up, the initiator would not become available again (signalled by LEDs on and flashing) until the animal had removed its nose for 1s. This ensured the animal would have a chance to focus on the goal-box cues before making a decision to go or re-initiate. Due to the failure of the nosepoke in one apparatus, an improvised nosepoke was set-up in one apparatus using a green LED connected to a breadboard and controlled using an Arduino microcontroller. Whereas no IR beam was present in the nosepoke, the experimenter could trigger the same LED sequence (off for 1 s, flashing for 4 s, then on) with the push of a button if the mouse performed a nosepoke.

General information and definition of responses
Horses. Horses were trained and tested in two batches by one experimenter (SH). Animals received one session per day. During habituation, and for part of shaping, the experimenter was present inside the arena, while remaining outside thereafter. The experimenter was then positioned behind the wall with the goal-boxes from where she operated the trial initiator, the goal-boxes, and the delivery of rewards. Rewards were in place in the goal-box when the goal-box lid was opened. Horses were not food restricted but received their regular allowance of feed three times a day. while an incorrect Go response was recorded at the Left-Right Discrimination Stage when the animal's head or forepaws touched any closed goal-box first. This was to prevent the rats from screening the boxes for the presence of reward before taking a decision. As incorrect Go responses almost never occurred after this stage, this response was not recorded in further stages. A No-go response was recorded when the animal re-initiated or did not make contact with any goal-box within the maximum trial time.
Mice. Mice were trained and tested in two batches in three, virtually identical apparatuses by four experimenters in Batch 1 (EM, JDB, MB-S, SH) and three experimenters in Batch 2 (EM, JDB, MB-S). In the majority of cases, individual mice were trained and tested by the same experimenter. However, for mice in Batch 1, one experimenter was replaced (MB-S by SH) towards the end of the training period due to an emergency leave of absence, and for Batch 2, mice from one experimenter (EM) were shaped for two sessions by the other two experimenters (JB and MB-S) due to illness.
Up to three mice were tested or trained in the same room simultaneously. Due to the design of the larger experiment (Bailoo, Murphy et al., manuscript in preparation), only one mouse per cage was trained and tested in the JBT. Throughout the study, the experimenters were positioned behind the wall with the goal-boxes, from where they operated the goal-boxes and the delivery of rewards.
Rewards were only delivered after the animal had made a Go response. Mice were fed a restricted diet of standard rodent food (KLIBA NAFAG #3430, Switzerland) to maintain 85-90 % (+ 1-3 %) of their free-feeding weight.
Go and No-go responses were recorded by the experimenter from the Left-Right Discrimination stage onwards. A Go response was recorded if the animal placed its nose within the circumference of the goal-box hole. Since the rewards were only delivered after this response was made and due to the design of the apparatus (no physical boxes present), mice were unlikely to investigate closed goalboxes; incorrect Go responses were thus not recorded. A No-go response was recorded if the animal re-initiated or did not perform a Go response within the maximum trial time.

Training Stage 1: Habituation
Horses. Horses were used to regular handling prior to this study and no habituation to the reward was necessary, as it was their usual food. Only the positive goal-box was used at this stage. Habituation was deemed successful if the horse stayed calm inside the arena (no whinnying, no fast movements in the arena, no repeated defecation) when the goal-box was opened and closed, and when it readily fed from the positive goal-box.
Rats. Rats were extensively handled in the previous study (Lampe et al., manuscript in preparation), and by the experimenter in the present study before training began. Further, rats were habituated to the rewards in their home cage prior to training. Rats were initially habituated to the test arena in pairs (assigned to the same positive location), and then individually until no obvious signs of fear or anxiety were observed (freezing, defecation, excessive grooming). Only the positive goal-box was used at this stage. Initially the goal-box was open and rewards were freely available within and on the floor to encourage exploration. Once rats were readily eating from the goal-box, the lid was closed after each reward was obtained and only opened again when the rat was within one body length from the goal-box and facing towards it. Habituation was deemed successful provided rats were consistently moving away and returning to the goal-box when it opened to obtain the reward.
Mice. Mice were extensively handled by three experimenters (EM, JB, MB-S) in their home cage and were habituated to the rewards in their home cage and to the test arena (though without the nosepoke and goal-boxes) prior to this study. Only the positive goal-box was used at this stage. As they were already habituated to the test arena, they were only given one 10-min habituation session in the arena where rewards were initially freely available in the goal-box. Once a mouse had obtained a few of the freely available rewards, rewards were only delivered once the mouse placed its nose within the circumference of the hole in the goal-box. Next, the goal-box door was closed and only opened again when the mouse was close to the goal-box and moving towards it. Mice which failed to eat within this first session were given a second habituation session following the same procedure until all mice were consistently moving away and returning to the goal-box when it opened to obtain the reward.

Stage 2: Shaping for Trial Initiation
Horses. Horses were shaped for trial initiation using the backwards-chaining method (McGreevy and Boakes, 2007), similar to Neave and colleagues (2013). Only the positive goal-box was used at this stage. Horses were trained to first touch the initiator (i.e. the plastic bottle), held by the experimenter to receive a reward (also held by the experimenter). Once a horse had learned to touch the initiator, the initiator was placed in its final position at the back of the arena. The experimenter then gradually moved away from the initiator towards the goal-boxes so that the horse had to move back and forth between the initiator and the experimenter to obtain the reward.  testing. Due to an experimenter's error, Block A in horses was comprised of different trials compared to rats and mice (see below), however this arrangement was decided a priori (i.e. not based on the performance of the horses) and was not deemed to influence the results, considering that the same strict performance criterion was applied across the three species.

Testing Stage 5: Judgement Bias Testing
Horses. All goal-boxes were used in this stage; the positive and ambiguous goal-boxes were always rewarded. In six sessions of 53 trials, where the maximum trial time was 10 s, Go and No-go responses were recorded per trial. Ambiguous trials were presented in trials 16, 32, and 48.
Rats. All goal-boxes were used in this stage; the positive and ambiguous goal-boxes were always rewarded. In six sessions of 53 trials, where the maximum trial time was 5 s, Go and No-go responses were recorded per trial. Ambiguous trials were presented in trials 16, 32, and 48.
Mice. All goal-boxes were used in this stage; the positive and ambiguous goal-boxes were always rewarded. In six sessions of 53 trials, where the maximum trial time was 5 s, Go and No-go responses were recorded per trial. Ambiguous trials were presented in trials 16, 32, and 48.