Social cues may prime social responses in learners that lead to deeper cognitive processing during learning and hence better test performance. The personalization principle is that people learn more deeply when the words in a multimedia presentation are in conversational style rather than formal style. This principle was supported in 14 out of 17 experimental tests, yielding a median effect size of d = 0.79. Some important boundary conditions are that the personalization principle may not apply to high-achieving students or long lessons. The voice principle is that people learn more deeply when the words in a multimedia message are spoken in a human voice rather than in a machine voice. This principle was supported in 5 out of 6 experimental comparisons, with a median effect size of d = 0.74. A possible boundary condition is that the voice principle may not apply when there are negative social cues such as low embodiment. The image principle is that people do not necessarily learn more deeply from a multimedia presentation when the speaker’s image is on the screen rather than not on the screen. This principle is based on 14 experimental tests in which half produced negative or negligible effects, yielding a median effect size of d = 0.20. The embodiment principle is that people learn more deeply when on-screen agents display humanlike gesturing, movement, eye contact, and facial expressions. In 11 out of 11 experimental comparisons, people performed better on transfer tests when they learned from a high-embodied agent than from a low-embodied agent, yielding a median effect size of d = 0.36. A possible boundary condition is that the embodiment principle may not apply when there are negative social cues such as a machine voice.