The overall aim of this project is stated to be to “determine how behavioural flexibility works and how we can make predictions about a species' ability to adapt their behavior to new environments”. Its aims are “to manipulate grackle behavioral flexibility … to determine whether their flexibility is generalizable across contexts, whether it is repeatable within individuals and across contexts, and what learning strategies they employ. Although I find these specific questions divorced from the background of invasive spread, and hence, do not particularly buy into the advantage of using an invasive species undergoing a current range expansion (in fact, I explain below how I find some predictions inherently contradictory with this background), the questions are worthwhile asking. Detailed comments are as follows: Please clarify: “they generally get faster rather than getting faster with each reversal” and how the learning criterion was altered to accommodate this

P1: Although I find it interesting to compare the extent to which rule-learning generalizes across contexts as a function of experience with rule learning, the question seems to contradict the notion that behavioural flexibility (measured here as rule learning) can be a target of selection during a range expansion, meaning that individuals would need to differ in some inherent way in their capacity to rule learn. Hence, for behavioural flexibility to be a target of selection, behavioural flexibility should not be flexible, if that makes sense. Put differently, invasive birds should not become broader-context rule-learners as a consequence of experience with rule learning. In fact, learning this capacity would be predicted to delay selection on the trait. Hence, H3a seems the more logical prediction and test, given the larger context of the project.

P1 alternative: “If the number of trials to reverse a preference does not correlate with or positively correlates with reversal number”; this contradicts the general pattern of finding of an SDR (number of trials to criterion decreases with increased reversal number) and also the authors’ own statement under P1: “Individuals improve their flexibility on a serial reversal learning task using colored tubes by generally requiring fewer trials to reverse a preference as the number of reversals increases”. Is this a typo?

Methods:

Learning criterion: is the sequence re-started at each testing session? Ie what is planned if a bird is willing to do 10 choices on one day and then satiates. Is the trial number started again at each testing session? Or do the, for example, correct trials yesterday count towards today’s? ie does the sliding window for scoring carry across testing sessions?

IVs: What is ‘batch’? test cohort? How many birds are tested simultaneously?

“we should expect due to the lack of data on this species for these experiments”: ok, but there are several published comparative data sets on SDR that you could use (e.g. Tebbich et al.).

“Analysis”: N=32: does that mean 16 experimental and 16 controls?

In measuring the latencies to switch to new options on the multi-access box: how can ‘work-time’ be disentangled from ‘non-work time’ in a latency? I.e. a bird that tries lots of options and interacts with the box might have the same latency as a bird that does mostly other things but then tries once and by chance solves the box immediately. How can these two seemingly very different birds be differentiated in a latency? I guess this is equivalent to asking whether multi-access box testing yields any motivation variables, similar to number of beak-to-task contacts one can obtain from an innovative problem-solving task. I would predict that motivation variables, which end up providing some measure of persistence, will be associated with exploitation vs. exploration strategies. I could not find an answer to this question in the protocol.

“Learning strategies”: The authors might find the following research and analysis strategy useful as it disentangles exploration from exploitation phases within each reversal. Federspiel, I. G., Garland, A., Guez, D., Bugnyar, T., Healy, S. D., Güntürkün, O., & Griffin, A. S. (2017). Adjusting foraging strategies: a comparison of rural and urban common mynas (Acridotheres tristis). Animal Cognition, 20, 65–74. Protocol_flexibility:

Why will birds only undergo 5 reversals on the touch screen? Is this enough to establish a significant slope in learning speed that can be compared across tasks?

I am also confused by not testing the control group. Was the suggestion not to train birds on the tube task (experimental and control) and then to compare the transfer of both these groups to each of the two additional tasks (multi-access and touch screen)? Why would the controls not participate in the reversals?

Reviewed by Andrea S. Griffin