This study was approved by the Ethics and Animal Welfare Committee of the University of Veterinary Medicine Vienna in accordance with good scientific practice guidelines and national legislation. The experiment was purely appetitive and strictly non-invasive and was therefore classified as a non-animal experiment in accordance with the Austrian Animal Experiments Act (TVG 2012). The part of the experiment conducted in Indonesia was approved by the Ministry of Research, Technology and Higher Education (RISTEK) based on a meeting by the Foreign Researcher Permit Coordinating Team (10/TKPIPA/E5/Dit.KI/X/2016) who granted the permits to conduct this research to M.O. (410/SIP/FRP/E5/Dit.KI/XII/2016) and B.M. (411/SIP/FRP/ E5/Dit.KI/XII/2016).

1. Preconditions/prerequisites

1. Basics
   1. Ensure that subjects can be identified individually. The study species may have distinct individual patterns or individuals can be marked (e.g., with color rings or non-toxic paint).
      NOTE: For more information on ringing as well as catching and releasing wild Goffins see Capture-release Procedure in Supplementary Information of³².
   2. Ensure a visually occluded room is available for testing to avoid social learning between subjects.
   3. Identify a highly preferred reward for the study species and group by testing multiple different, available treats (see³³ or Food Preference Test in Supplementary Information of reference³²).
   4. Consider whether feeding time substantially differs between the groups. If so, consider a protocol that ensures the feeding time does not heavily reduce the time available to solve tasks for one of the groups (see step 4.8 for more information).
      NOTE: In this study, there was a preference of the long-term captive group for cashews and for dried corn in short-term captive group.
2. Designing the Innovation Arena
   NOTE: The full apparatus, i.e., the Innovation Arena, consists of 20 different puzzle boxes, arranged in a semi-circle on a wooden platform.
   1. Design the basic outline of the boxes in a size applicable for the study species. Use transparent boxes having trapezoid shape (for easy alignment in a semi-circle), removable lids (to allow baiting in between sessions), and detachable bases (see Figure 1).
      NOTE: Each base will later stay in a permanent position while the rest of the boxes will change positions. In the presented study, the size of the boxes was chosen to assure that each puzzle is easily accessible by the cockatoos. The dimensions can be adjusted for each study species.
   2. Design a platform to hold the 20 puzzle boxes.
   3. Design a fixation system that will keep the lids of the boxes in place during testing and, therefore, cannot be removed by the subjects during test sessions.
      NOTE: It has to be detachable from the apparatus as the lids of the boxes need to be taken off for baiting.
   4. For the front of each box, design 20 different tasks, each of which will constitute a different technical challenge (see Figure 2).
      NOTE: The tasks for this experiment were designed with the aim that solutions fall within the morphological range of many different species. For comparative strength, it would be ideal to use tasks as similar to these as possible but keep in mind that it is of even greater importance that the tasks are novel to the subjects. See the Table of Materials for exact measurements and the Supplementary Technical Drawing for a more detailed illustration of the tasks.
   5. Acquire all material needed for the apparatus.
   6. Ensure to have a wide-angle camera, a coding software (recommended, e.g., Behavioral Observation Research Interactive Software, BORIS³⁴) and software for statistical analysis (recommended, e.g., R³⁵).
      NOTE: For field studies, ideally, design the arena before leaving for the study site and bring as much as possible of the essential equipment, such as pre-cut acrylic glass, along.

Figure 1: Diagram of a basic three-sided box. Please click here to view a larger version of this figure.

Figure 2: Tasks of the Innovation Arena with a corresponding description of the motoric action required for solving (  = reward; red arrows indicate directions of actions required to solve tasks; yellow arrows indicate reward trajectories). Tasks are arranged according to their mean difficulty (left to right, top to bottom). Previously published in³². Please click here to view a larger version of this figure.

2. Preparations

1. Glue together three sides of the boxes: left, back, and right side but not the front, top and base.
2. Position each three-sided box on top of each base and evenly align them in a semicircle on the platform (Figure 3). The front section of each box should sit 1 m from the center.
   NOTE: The mechanisms constituting the task (front faces of boxes and possible contents) will be added at a later point during the experiment.
3. Draw a line from each box 20 cm toward the center of the arena and connect the lines, resulting in a proximity grid (Figure 3).
   NOTE: Depending on the size of the study species, a different distance might be more appropriate. For the study presented here, 20 cm was chosen as it is roughly the length of a Goffin (tail feathers excluded).
4. Remove everything except the bases of the boxes and attach them permanently to the platform. This will ensure that the boxes will stay in place during the experiment.
5. Attach a wide-angle camera on the ceiling above the arena.
6. Prepare a schedule for the position of each box per session and subject. Each subject will always be confronted with all boxes, but with a new arrangement each session. The location (positions 1 to 20) of each task should be randomly assigned with the restriction of no box being at the same position twice per subject.
   NOTE: This is the ideal situation. If one cannot plan the test order of subjects (which is more likely in field studies) this randomization limitation (no box at the same position twice) between sessions (but not within subject) must suffice.

Figure 3: The Innovation Arena. Tasks arranged in a semi-circle; the positions of the 20 tasks are exchangeable. A proximity grid (20 cm in front of each box) is marked in black. Please click here to view a larger version of this figure.

3. Habituation

NOTE: The purpose of habituation is to reduce influences of neophobic reactions toward the arena. Ensure a minimum habituation level for all subjects through a habituation procedure that requires each individual to reach two criteria.

1. Habituation to non-functional arena (until criterion I)
   1. Position all three-sided boxes on the bases, add the lid of each box and hold them in place with the fixation system (without the subject present).
      NOTE: Consider habituating the subjects in stages that are appropriate for each species, for example, by incrementally adding more boxes to the platform, presenting the arena in their home area, placing rewards at any position of the platform such as around, on top, and with the boxes or confront them with the apparatus in bigger groups first and gradually minimize the group size.
   2. Familiarize subjects with separate elements of the tasks that might elicit neophobic reactions.
      NOTE: These separate elements (i.e., everything but the basic boxes, platform, and fixation system) must not be combined into functional mechanisms at this stage.
   3. Place one reward inside the box (center). Bring the subject into the compartment.
   4. Wait for the length of a session without interfering. Subjects are now supposed to eat the rewards.
      NOTE: The duration of these habituation sessions differed in the experiment: long-term captive birds received 10 min, while short-term captive cockatoos had 20 min to eat the rewards. This was necessary to account for a substantially longer feeding time due to different reward types. This issue was addressed differently later in the test sessions (see step 4.8).
   5. Repeat for each subject (one session per test day) until criterion is reached: Each individual consumes all the rewards from three-sided boxes (one reward per box) within three consecutive sessions while being visually isolated from the group.
2. Habituation to functional arena (until criterion II)
   1. Glue and permanently attach all the necessary elements to the boxes to make them functional puzzle boxes.
      NOTE: At this point, the arena is fully functional as in the test sessions.
   2. Place the boxes randomly on the platform (they will be kept in place by the bases) and secure the lids to the boxes.
   3. Place one reward on the lid of each box on the edge closest to the center of the arena.
   4. Bring the subject into the compartment.
   5. Wait for the length of a session without interfering.
      NOTE: Subjects are supposed to now eat the rewards.
   6. Repeat for each subject (one session per test day) until the criterion is reached.
      NOTE: Criterion II: Individual consumes all the rewards from the top of the functional puzzle boxes (one reward per box) within one session while being visually isolated from the group. This criterion II will ensure that the subjects are not afraid of the arena, even when new parts are attached. They should however not interact with the mechanisms and should be interrupted if they do so.

4. Testing

1. Place the boxes on the platform according to the randomization schedule.
2. Bait each task at the appropriate location inside the boxes (see Figure 2).
   NOTE: The exact location of each reward depends on the specific task and can be seen in the video.
3. Attach the lids to the boxes and secure them with the fixation system (to make sure subjects cannot pull them off).
4. Separate one individual subject and bring it into the test compartment. Subjects are tested one at a time to avoid interference of social learning.
5. Either position them on the start position (i.e., the point which is at equal distance to all the tasks at the center of the platform) or place an incentive (e.g., a reward) at the start position to ensure that the subject begins there.
6. Start the timer and wait for 20 min (session duration) without interfering or interacting with the subject. The subject can solve as many tasks as possible.
7. If the subject gets distracted with non-apparatus related objects, the experimenter is allowed to place them back at the start position of the arena (if possible).
8. If the subject feeds for longer than 3 s on the reward, stop the timer, wait until the feeding is finished, and then resume timing.
   NOTE: This is done to ensure that the maximum time available to solve tasks is not reduced by feeding time and therefore equal for both groups.
9. If the subject does not interact with any task within the first 3 min and is also not agitated, apply a motivation protocol (see section 5).
10. Once the 20 min have elapsed (maximum duration of one session) or the participant has solved all the tasks, the subject is done with the testing for the day and can be released back into the home area.
11. On the next test day, repeat this procedure.
12. Continue testing each individual until it either does not solve any new task in the last five sessions or does not solve any task at all in 10 consecutive sessions.

5. Motivation protocol

NOTE: As described above (step 4.9), a motivational protocol can be implemented if an individual does not interact with any task within the first 3 min of a session.

1. Place three rewards on top of the boxes (choose a box on the left, middle, and right side for this). If the subject starts interacting with any task 3 min after consuming the rewards, resume the session (the 20 min duration starts at this point).
2. If not, place five rewards dispersed on the approach line (i.e., proximity grid). If the subject starts interacting with any task 3 min after consuming the rewards, resume the session (the 20 min duration starts at this point).
3. If not, place five rewards at the starting position. If the subject starts interacting with any task 3 min after consuming the rewards, resume the session (the 20 min time frame starts at this point).
4. If not, place a handful of rewards at the start position and terminate the test session for this day (but give the subject some time to consume the rewards).

6. Analysis

1. Behavioral coding
   1. Before analyzing the videos examine the coding protocol in detail (Table 1) and consider whether adjustments are necessary for the species being tested.
      NOTE: The descriptions of the coding variables should be as specific as possible in order to avoid coding differences between various researchers.
   2. Annotate point events of: Number of different tasks touched (TasksTouched; Note that the maximum number of tasks touched is 20), number of tasks solved (TasksSolved), contact with baited tasks (BaitedContact), and contact with solved tasks (SolvedContact).
   3. Annotate durations for latency until the subject crosses the outer boarder of the grid line (LatencyGrid) and time spent within the grid (GridTime).
2. Statistical analysis
   1. Determine whether measures for apparatus-directed behaviors (LatencyGrid, GridTime, TasksTouched, BaitedContact, SolvedContact) are correlated.
   2. If yes, then extract the principal components using a Principal Component Analysis before including them in the model as predictors.
   3. If they are not correlated, include them separately in the model as predictors.
   4. Run a Generalized Linear Mixed Model with binomial error structure and logit link function³⁶. To predict the probability of success (i.e., the response variable being SolvedTasks), fit the model with maximal random slope structure and include random intercepts for subject, task, a combined factor of subject and session (SessionID), and a combined factor of subject and task (Subj.Task) to avoid pseudo-replication. Use the comparison of interest (e.g., species) and the Principal Components as predictor variables and control for session. Consider possible interactions.
   5. To avoid cryptic multiple testing³⁷ first compare the model with a model lacking all fixed effects of interest before testing individual predictors.
   6. To test for an overall difference of difficulty in tasks between groups, compare the (full) model with one lacking the random slope of group within task.