In this talk I will review some of the highlights from the experimental pragmatics literature while arguing for a distinction between two kinds of pragmatic enrichments, those that I call voluntary and those that I call coerced. Voluntary pragmatic enrichments, exemplified by scalar inferences, are optional in the sense that the linguistically encoded meaning of an utterance could be good enough for discerning a speaker's informative intention; in these cases, enrichments arise in order to produce a narrower interpretation. That is, while Some cats are mammals can be considered true if one employs Some with its lexical meaning (which can be viewed as good enough for processing), a listener who has the processing capability is in the position to enrich the meaning of Some (so that it is taken to mean Some and not all) and produce a false response. Coerced readings, exemplified by metaphor, arise when a pragmatic enrichment is practically necessary on the part of the listener in order to arrive at a reasonable hypothesis about the speaker's meaning. Sentences such as My son is a tadpole would be false without such effort. This distinction between voluntary and coerced enrichments can be handy for describing empirically difficult cases such as the invited inferences linked to conditionals (e.g. how if p then q appears to generate if q then p) and for describing why data appear to vary with respect to those who fall on the autism spectrum.