High-level cognition is often accomplished not byindividuals working in isolation, but by distributed, complexcognitive systems. Examples include teams of scientists orcollaboratively improvising musicians. These distributedsystems can undergo critical transitions, suddenly movingfrom one stable pattern of activity to another. For instance,in ‘free jazz,’ where musicians improvise without apredetermined plan or a central leader, the performance willoften settle into a particular texture or style beforetransitioning to something entirely new, often quitesuddenly. When do these transitions occur? Are theyforeseeable? Inspired by suggestions that cognitive systemsare, in some sense, a kind of ‘ecosystem,’ we draw on recentwork in quantitative ecology that has begun to describegeneric early warning signals of impending criticaltransitions in ecosystems. We apply these techniques to acorpus of audio recordings of professional jazz quartetsplaying improvised music. We find that the same genericmeasures that have been used successfully to predict criticaltransitions in natural ecosystems describe the complexdynamics of improvised musical performance in the lead-upto transitions. By taking seriously the metaphor thatcognition occurs in ‘ecosystems,’ we gain new insights intohow stable patterns of thought can emerge suddenly incomplex cognitive systems.