The current state of psychological science is such that we frequently fail to achieve one of the major goals of research: to generalize findings beyond the samples, stimuli, and paradigms on which they were observed. Generalizations are constrained by restricted and nonrepresentative samples, biased operationalizations and overbroad characterizations, stimulus and paradigmatic threats to construct and ecological validity, and overreliance on mean differences at the expense of attention to individual variability. I illustrate these threats to generalization using examples from cognitive research. I contend that these practices lessen the authenticity of our research and if left unchecked, will render our discipline irrelevant and obsolete, a threat as existential as that posed by practices that limit rigor and replicability. I make suggestions for ways to improve current research practices that will help to build a more valid, generalizable, and applicable science, thus advancing both basic and applied research in memory and cognition.