8.6 How inferences arise, and neurodiversity in inference making

One of the underlying assumptions in the Cooperative Principle is that in all conversations, people have good faith in their conversation partners: everyone assumes that everyone is being rational and cooperative, except in special circumstances. This assumption is the foundation of how implicatures are created in discourse. Let’s revisit the discourse from earlier.

The inference that Aya would naturally make from Bo’s response is that Raj did NOT clean the litterbox. Aya arrives at this inference because she assumes Bo is following all four maxims in the conversation. Particularly, Aya’s faith in the Maxim of Quantity leads to the conclusion that this must’ve been the most informative thing Bo could’ve said.

Here are more examples with the other three maxims.

According to the Cooperative Principle, the default assumption that Aya would make after Bo’s utterance in (2) is that Bo has some way of knowing The Blue Jays won is true, whether it be direct evidence (e.g., he was at the game) or indirect evidence (e.g., he saw it on the news), and that the evidence is reliable. This is based on the Maxim of Quality, which states that discourse participants should not utter things that they don’t have sufficient evidence for. Depending on the context, you may get various implicatures about how they know what they have said to be true. For example, if Aya knows that Bo was in class during the game, she might infer that Bo saw the results of the game online.

In (3), the Maxim of Relation plays a significant role in the creation of the implicature.

Here, Bo’s utterance at first glance may seem unrelated to the question that A posed. To reiterate how the Cooperative Principle works, the idea is that Aya does not automatically assume that Bo is being uncooperative when something like this happens. The default assumption that Aya would make is that Bo is being cooperative and following all four maxims. Based on this, Aya would make the calculation that Bo’s utterance is somehow related to the question she posed. For example, the relation she might infer may be that vacuuming creates a lot of noise, the noise would wake the cat up, and he didn’t want to do that. Hence, ‘No, Bo didn’t vacuum’ results as the inference. Remember the chicken nuggets example in Section 8.5 (Aya: What extracurriculars did you do? / Bo: ??My favorite food was chicken nuggets.)? There, if you found yourself trying to make a connection between Bo’s utterance and Aya’s question, the Gricean explanation is that this is because you want to believe that Bo is following the Maxim of Relation. This kind of faith is what the Cooperative Principle is all about.

Here is a question for you (the reader/listener/viewer): in an example like (3), did you actually draw the same inference as Aya? Were any of you uncertain about whether that kind of inference could be drawn from the given utterances? If you did not draw the same inference as Aya or if you were not super confident about making the inference Aya made, that’s totally OK! We mentioned already that conversational rules can vary from language to language and from culture to culture — in addition to that, there is also quite a bit of individual variation when it comes to drawing inferences in a conversation. The inference in (3), for example, depends on things like how much and what kind of experience you have with vacuums and cats. Another factor in pragmatic variation involves autism. For example, studies have shown that autistic adults and non-autistic adults sometimes have different strategies for drawing inferences in a conversation.

In one study, researchers analysed how 66 autistic adults and 118 non-autistic adults interpreted conversations like the following (Wilson & Bishop 2020):

Participants in this study were asked whether they thought Character 2 heard what the police said or not. Participants were given three answer options: “Yes”, “No,” and “Don’t know”. Non-autistic adults typically answered “No” in this kind of context. In (4) in particular, this is an inference from Character 2’s utterance: lots of trains imply lots of noise, which implies “No, couldn’t hear what they said”.  Autistic adults were about 2.5 times more likely than non-autistic adults to answer “Yes” in a context like this. Autistic adults were also about 6 times more likely than non-autistic adults to answer “Don’t know” in a context like this. Importantly, autistic adults and non-autistic adults performed about the same in other linguistic tasks (e.g., vocabulary tests, syntactic acceptability judgment tasks, and comprehension of literal meanings in a conversation). This suggests that non-autistic adults and autistic adults’ pragmatic abilities in inferencing in particular were different.

In a follow-up study, the researchers repeated the same inferencing task with just the autistic participants. This time, they eliminated the “Don’t know” option to see what the autistic participants would do when forced to choose “Yes” or “No” in a context like (4). Results showed that if an autistic participant chose “Don’t Know” in the first task in a context like (4), they had about a 91% probability of choosing “No” in this second task. This suggests that when constrained to do so, autistic participants generally gave the same response as non-autistic participants in this study. Applied to everyday situations, this might mean that autistic adults can get the implicatures intended by non-autistic adults’ — but their tendency to do so is different.

Another interesting insight from this study is the comments that the autistic participants gave after the experiment: several autistic participants expressed frustration that there wasn’t enough information to answer the implicature questions. We might wonder if autistic adults and non-autistic adults have a difference in their Maxim of Quality (“Only say what you believe to be true, and only say what you have adequate evidence for”): could it be that what counts as “adequate evidence” is stricter for autistic individuals? Some autistic participants also made comments like this:

Based on this kind of comment, autistic adults’ tendency to select “Don’t know” in the first task might also be driven by their history of being criticised for not getting non-autistic adults’ intended implicatures.

What do we learn from all of this? What inferences can be made from an utterance depends on the context, and “context” includes who is participating in the discourse, too. The quote above points to yet another power dynamic in language (recall Chapter 2): conversational expectations are often very neurotypical-centric, which is unfair to neurodivergent individuals. Some people may not get the implied meaning immediately, and even when they think they understand the implicature, they might not feel confident about it. Following the Autistic Self-Advocacy Network‘s advice, we urge you to be patient if miscommunication arises in everyday situations. If you are organizing an event, you may want to avoid relying on implicatures to communicate important information to participants, because some people might not get it. Everyone has a different mind, so let’s support these differences rather than suppress them.

Check your understanding

References

Grice, H. P. (1975). Logic and conversation. In Speech Acts (pp. 41-58). Brill.

Hughes, J. M. (2016). Increasing neurodiversity in disability and social justice advocacy groups. Washington, DC: Autistic Self Advocacy Network.

Wilson, A. C., & Bishop, D. V. (2020). “Second guessing yourself all the time about what they really mean…”: Cognitive differences between autistic and non‐autistic adults in understanding implied meaning. Autism Research, 14(1), 93-101.