Functional neuroimaging has opened the door for examining in vivo neural processes for human perception, cognition, and behavior. Naturally, we wonder if we can ‘read the mind’ from brain images. Typical methods of fMRI are not apt for this, however, as it is aimed at answering ‘given a mental state, which brain region is active?’ rather than ‘given a brain activity, which mental state?’. As a solution, I examine a method using the entire brain to build ‘brain decoders’ that can empirically measure mental processes. Since different mental processes are unlikely to share the exact same pattern of whole-brain activity, whole-brain decoders could improve specificity to mental processes. Simultaneously, because it recruits multiple regions’ signals, whole-brain decoders could also improve our sensitivity to detect mental processes. In the first study, I address the statistical and substantive difficulties of whole-brain decoders and propose a novel algorithm that can overcome them. In the second study, I build a whole-brain decoder of valuation across two economic decision-making tasks and show how a post-hoc analysis of the decoder can yield insight into signal relationships between different regions. In the third study, I showcase how empirical measurements of mental processes can be used in psychological research. Existing theories have posited that people discount delayed rewards because they are imagined less vividly than immediate rewards. I provide neural evidence to this claim by building a whole-brain decoder of imagination vividness on one dataset and show that it can also predict temporal delay in two other datasets of delay discounting task. This dissertation, taken together, shows that whole-brain decoders can be an easy analysis to implement that captures unique neural signatures of tasks and provide measurements of mental processes and constructs.