What is essential to a model are the hypotheses about particular relations between distinct propositions that can be derived. It is hard if not impossible to test a model (let alone a theory) as a whole, and research typically focuses on one or two selected aspects. The main hypothesis for metaphor comprehension in DMT is that most metaphor in language is not processed as a metaphor in comprehension.

There are also hypotheses that have been proposed by other researchers that are compatible with DMT. For instance, there is the Grammatical Concordance Hypothesis advanced by Brian Bowdle and Dedre Gentner that says that there is an interaction between the grammatical form of metaphor in language and metaphor in processing. Also, there is the Graded Salience Hypothesis by Rachel Giora, which says that the nature of metaphor comprehension is determined by the distinction between salient and non-salient senses, not metaphorical versus non-metaphorical senses. This is related to another hypothesis, not labelled as such, that was formulated by Sam Glucksberg, who states that there is no priority for literal meaning in utterance processing. All of these hypotheses have been supported by experimental evidence and have motivated the formulation of DMT. They have now been incorporated within the model for DMT.

The combination of these hypotheses in DMT also gives rise to new ones. Thus, the Grammatical Concordance Hypothesis was limited to the distinction between metaphor and simile and their relation with processing by lexical disambiguation versus analogy.  However, there are other forms of metaphor than metaphor and simile, and DMT can advance and test related predictions for those too. Moreover, the four dimensional model also raises questions about relations to other dimensions than the one dealing with grammatical form, so that, for instance, hypotheses are facilitated about variation in reference and its relation to metaphor in processing. This in turn triggers hypotheses about deliberate metaphor and conscious metaphorical cognition, and intentions and resistance to metaphor. This is how DMT hypotheses can eventually feed back into the theory of DMT and account for the key issues of the paradox of metaphor, the ambiguity of metaphor, and the multivalence of metaphor.

Other hypotheses can arise through the reinterpretation of existing research. For instance, Valentina Cuccio has proposed that there is a meaningful distinction beween early versus late somatotopic activation in the brain in interaction with different types of metaphors. She has linked this with the emergence and role of simulation. This is a new theoretical proposal that fits with the model and invites dedicated experimental testing.

All of these hypotheses derive from the DMT model and can be operationalized in the same way for the metaphor materials by applying the formalism developed in Slowing metaphor down.