Understanding the conundrum of human cooperation has been declared one of the century’s grand challenges. Traditionally, the evolution of cooperative action in nature is analyzed through the lens of Evolutionary Game Theory, specifically, using the social learning framework, a model for Darwinian competition. However, more complex individuals may resort to more sophisticated learning rules such as Counterfactual Thinking (CT). Given these individuals’ cognitive empowerment, the question of how the presence of counterfactuals influences the evolution of cooperation in a hybrid population of these complex agents and social learners. Here we explore how cooperation emerges from the interplay of different strategy revision paradigms by analyzing large-scale Markov processes. We find that increasing the prevalence of CT individuals can promote cooperation, but such an increase is non-monotonous. Moreover, whereas counterfactual reasoning generally fosters cooperation, it fails to promote such behaviour among counterfactuals. Lastly, we find that increasing the population’s heterogeneity level enhances cooperation among social learners, but again not among counterfactuals. This indicates that, under certain circumstances, the presence of more sophisticated agents may help promote cooperation in hybrid populations. The proposed study may come as a starting point for a more profound understanding of agents’ counterfactual rationality impact on hybrid populations.