TABLE 5.

The claims as presented in Claims: Expected Observations that need to be justified in a co-adaptation experiment, including whether they were validated and an explanation of that conclusion.

Claim	Justified	Explanation
Different participants develop different ways of performing the task	Yes	When looking at the different interaction patterns that people engage in, and categorizations of their adaptive behavior, we can see that different people indeed performed the task in a variety of ways
The agent learns different sequences of strategy options for different participants	Partly	The results showed that not all agents learned the same model on an individual level. However, the models had much in common, suggesting that all agents learned similar behavior. When splitting this up in groups based on human adaptive behavior, there seems to be a difference in learned agent behavior between the different groups. Currently, however, we did not do any statistical analysis to test whether this is a significant result
Different teams converge to different ways of performing the task	Partly	When looking at the different interaction patterns that participants engaged in with their robot team partner, different teams solved the task in a variety of ways (see H1). However, it is unclear to what extent the robot contributed to this. Moreover, while participants generally gained more confidence in their strategy and expressed to experience a greater subjective collaboration fluency toward the end of the experiment, it is unclear to what extent the strategy of the team really converged to a stable one
The agent converges to a specific sequence of strategy options for most participants	No	While we did observe a logical development of the Q-values on a population level, this does not count for all of the individual agents. Moreover, it is not clear to what extent the agents really converged to a stable set of actions
The human converges to a specific strategy within the experiment	Partly	The categorizations of participant behavior show that participants settle on a stable strategy more and more over the course of the experiment. This is also shown by the development of the confidence scores and subjective collaboration fluency. True convergence to a stable strategy, however, is not clearly visible within the 8 runs of the experiment