Behavior change interventions (BCIs) have been used to improve infant and young child feeding (IYCF) practices with varying success. This may result from inadequate consideration of determinants of behavior, including "caregiver capabilities." We aimed to: 1) examine the extent to which caregiver capabilities are considered in research on complementary feeding BCIs in low- and middle-income countries, 2) describe IYCF trajectories from 0 to 11 months of age and explore caregiver decisions at critical IYCF junctures, and 3) examine the role of caregiver self-efficacy for complementary feeding as part of a program impact pathway to improved behaviors. We conducted a scoping study of the peer-reviewed complementary feeding BCI literature (objective 1); used ethnographic methods to collect and analyze in-depth qualitative longitudinal interviews from the process evaluation the Alive & Thrive BCI in Bangladesh (objective 2); and conducted structural equation modeling to test the direct and indirect effects of self-efficacy for two complementary feeding behaviors (objective 3) using survey data from a process evaluation of the Alive & iii Thrive BCI in Bangladesh. In the scoping study (objective 1), we found that caregiver capabilities are rarely mentioned, intervened on, or measured in BCI research on complementary feeding, revealing considerable gaps in this literature. In the study of IYCF trajectories (objective 2), we observed substantial intra-cultural diversity, resulting from a combination of child, caregiver, and household factors, suggesting no normative longitudinal patterns for IYCF in this study population. We identified consequential junctures in IYCF, "decision moments," that determined each child's IYCF trajectory. These findings indicate the value of individually tailored interventions to effectively target decision moments. The Alive & Thrive BCI improved two complementary feeding behaviors that we analyzed (objective 3). For one behavior, feeding green leafy vegetables, the BCI operated though self-efficacy and mothers with greater selfefficacy were more likely to practice this behavior. For the second behavior, on-time introduction of egg, the BCI did not work through self-efficacy, likely due to women's lack of resources, autonomy, and access to markets. The use of multiple methods advanced our understanding of intervention pathways and highlighted the important roles of caregiver capabilities in this context. iv