Even after research-based instruction, students who demonstrate the ability to assemble relevant conceptual knowledge on one physics question may have difficulty assembling that same knowledge on a closely related problem. Recent research has suggested that reflexive, bottom-up reasoning processes seemingly unrelated to the physics concepts themselves may be responsible for these difficulties. Research has also suggested that attending to these reflexive processes during instruction may improve performance to a greater degree than attending solely to top-down, reflective thinking. Leveraging these findings to meaningfully improve instruction is important. We have, therefore, investigated the impact of training focused on Newton's second law targeted at reflexive reasoning processes and compared results to a more standard reflective approach to the same topic. We find that an approach targeted toward reflexive reasoning processes improves performance on a difficult physics question to the same or greater degree as a typical reflective approach. Furthermore, we find that many students whose performance on a difficult physics question increased after the reflexive training also explained correct conceptual reasoning on that question, suggesting that conceptual understanding was bolstered by the bottom-up, reflexive training.