A dominant view in current memory research is that there are distinct implicit (unconscious) and explicit (conscious) memory systems. The present thesis proposes an alternative, single-system signal-detection model of priming (a traditional implicit memory phenomenon) and recognition (a traditional explicit memory phenomenon). The model has two core assumptions: 1) priming and recognition are driven by the same memory strength signal, and 2) this signal is subjected to independent sources of random noise for priming and recognition tasks (the variance of which is typically greater for priming tasks). The model is shown to account for numerous results: 1) the sensitivity of priming tasks does not typically exceed that of recognition tasks, and priming therefore does not occur when recognition is at chance (Experiments 1-8); 2) the magnitude of the effect produced by manipulations of attention at encoding is greater on recognition than priming (Experiments 5-8), and this can give rise to single dissociations (Appendix 1); 3) priming and recognition can be very weakly correlated, even though they are driven by the same memory signal; 4) priming can occur for items that are not recognised (Experiment 9; Simulation Study 2); 5) the relationship between the identification latencies to misses and false alarms can change as a function of overall memory strength (Simulation Study 3); 6) priming and fluency are relatively intact in amnesics, despite severe impairments in recognition (Simulation Study 4). Thus, contrary to previous interpretation, (2)-(6) are not inconsistent with a single-system view; (1) suggests that the contents of the memory driving priming are accessible to consciousness. Finally, the predictions of the model were tested in a novel paradigm, the CID-2AFC task (Experiments 10-12; Simulation Study 5). Limitations of the model (and a dual-system version) were revealed, suggesting directions for future research.