Can technologies like reCAPTCHA use the solver behavior to uniquely identify them?

It seems to me that the users' interactions with reCAPTCHA problems may be used to uniquely identify them from the response times, choice of image sequences or other behavior of users' interaction with the reCAPTCHA assuming that the sample of solved recaptchas that the user solves is large enough.

Is there any research on this(e.g experiments with reCAPTCHA or similar systems) if people can leave unique interaction patterns(and how unique these are between people) on these systems that could be …

identify privacy recaptcha technologies