Similarities and differences in understanding negative and affirmative counterfactuals and causal assertions: Evidence from eye-tracking

Mental simulation of the factual and the illusory in negation processing: evidence from anticipatory eye movements on a blank screen

How do comprehenders process negative statements such as The fish is not jumping out of the water? Opinions vary. Some argue for two steps, namely that processing starts off with the representation of the positive/illusory [fish jumping out of the water] and then shifts to the (f)actual. To test this idea, we measured fixations on the factual (fish not jumping) versus the illusory (fish jumping) during auditory processing of negation and affirmation. We tested speakers of English (single-cued negation) and Croatian (double-cued negation) and focused on anticipatory fixations in the absence of pictures to indicate the strength of mental simulations. Our findings contribute to negation processing research in two ways. First, dominant anticipatory fixations on the factual suggest a direct rather than a two-step process. Second, time-sensitive insights from two languages call for a finer-grained account of negation processing with negation-specific support of inferences of the factual over the illusory.

How people think about the truth of hypothetical impossibilities

People can think about hypothetical impossibilities and a curious observation is that some impossible conditionals seem true and others do not. Four experiments test the proposal that people think about impossibilities just as they do possibilities, by attempting to construct a consistent simulation of the impossible conjecture with its suggested outcome, informed by their knowledge of the real world. The results show that participants judge some impossible conditionals true with one outcome, for example, "if people were made of steel, they would not bruise easily" and false with the opposite outcome, "if people were made of steel they would bruise easily", and others false with either outcome, for example, "if houses were made of spaghetti, their engines would (not) be noisy". However, they can sometimes judge impossible conditionals true with either outcome, for example, "if Plato were identical to Socrates, he would (not) have a small nose", or "if sheep and wolves were alike, they would (not) eat grass". The results were observed for judgments about what could be true (Experiments 1 and 4), judgments of degrees of truth (Experiment 2), and judgments of what is true (Experiment 3). The results rule out the idea that people evaluate the truth of a hypothetical impossibility by relying on cognitive processes that compare the probability of each conditional to its counterpart with the opposite outcome.

How people reason with counterfactual and causal explanations for Artificial Intelligence decisions in familiar and unfamiliar domains

Few empirical studies have examined how people understand counterfactual explanations for other people's decisions, for example, "if you had asked for a lower amount, your loan application would have been approved". Yet many current Artificial Intelligence (AI) decision support systems rely on counterfactual explanations to improve human understanding and trust. We compared counterfactual explanations to causal ones, i.e., "because you asked for a high amount, your loan application was not approved", for an AI's decisions in a familiar domain (alcohol and driving) and an unfamiliar one (chemical safety) in four experiments (n = 731). Participants were shown inputs to an AI system, its decisions, and an explanation for each decision; they attempted to predict the AI's decisions, or to make their own decisions. Participants judged counterfactual explanations more helpful than causal ones, but counterfactuals did not improve the accuracy of their predictions of the AI's decisions more than causals (Experiment 1). However, counterfactuals improved the accuracy of participants' own decisions more than causals (Experiment 2). When the AI's decisions were correct (Experiments 1 and 2), participants considered explanations more helpful and made more accurate judgements in the familiar domain than in the unfamiliar one; but when the AI's decisions were incorrect, they considered explanations less helpful and made fewer accurate judgements in the familiar domain than the unfamiliar one, whether they predicted the AI's decisions (Experiment 3a) or made their own decisions (Experiment 3b). The results corroborate the proposal that counterfactuals provide richer information than causals, because their mental representation includes more possibilities.

"Looking at" Negation: Faster Processing for Symbolic Rather Than Iconic Representations

Many studies have shown the double processing of negation, suggesting that negation integration into sentence meaning is delayed. This contrasts with some researches that have found that such integration is rather immediate. The present study contributes to this debate. Affirmative and negative compound sentences (e.g., "because he was not hungry, he did not order a salad") were presented orally in a visual world paradigm while four printed words were on the screen: salad, no salad, soup, and no soup. The eye-tracking data showed two different fixation patterns for negative causal assertions, which are linked to differences in the representation and inferential demands. One indicates that negation is integrated immediately, as people look at the explicit negation (e.g., no salad) very early. The other, in which people look at the alternate (e.g., soup) much later, indicates that what is delayed in time is the representation of the alternate. These results support theories that combine iconic and symbolic representations, such as the model theory.