People regularly form expectations about their future, and whether those expectations are positive or negative can have important consequences. So, what determines the valence of people’s expectations? Research seeking to answer this question by using an individual-differences approach has established that trait biases in optimistic/pessimistic self-beliefs and, more recently, trait biases in behavioral tendencies to weight one’s past positive versus negative experiences more heavily each predict the valence of people’s typical expectations. However, these two biases do not correlate, suggesting limits on a purely individual-differences approach to predicting people’s expectations. We hypothesize that, because these two biases appear to operate via distinct processes (with self-beliefs operating top-down and valence weighting bias operating bottom-up), to predict a person’s expectations on a given occasion, it is also critical to consider situational factors influencing processing style. To test this hypothesis, we investigated how an integral part of future thinking that influences processing style—mental imagery—determines each bias’s influence. Two experiments measured valence weighting biases and optimistic/pessimistic self-beliefs, then manipulated whether participants formed expectations using their own first-person visual perspective (which facilitates bottom-up processes) or an external third-person visual perspective (which facilitates top-down processes). Expectations corresponded more with valence weighting biases from the first-person (vs. third-person) but more with self-beliefs from the third-person (vs. first-person). Two additional experiments manipulated valence weighting bias, demonstrating its causal role in shaping expectations (and behaviors) with first-person, but not third-person, imagery. These results suggest the two biases operate via distinct processes, holding implications for interventions to increase optimism.