Examining the cognitive costs of counterfactual language comprehension: Evidence from ERPs

Abstract

Recent empirical research suggests that understanding a counterfactual event (e.g. 'If Josie had revised, she would have passed her exams') activates mental representations of both the factual and counterfactual versions of events. However, it remains unclear when readers switch between these models during comprehension, and whether representing multiple 'worlds' is cognitively effortful. This paper reports two ERP studies where participants read contexts that set up a factual or counterfactual scenario, followed by a second sentence describing a consequence of this event.Critically, this sentence included a noun that was either consistent or inconsistent with the preceding context, and either included a modal verb to indicate reference to the counterfactualworld or not (thus referring to the factual-world). Experiment 2 used adapted versions of the materials used in Experiment 1 to examine the degree to which representing multiple versions of a counterfactual situation makes heavy demands on cognitive resources by measuring individuals' working memory capacity. Results showed that when reference to the counterfactual-world was maintained by the ongoing discourse, readers correctly interpreted events according to the counterfactual-world (i.e. showed larger N400 for inconsistent than consistent words). In contrast, when cues referred back to the factual-world, readers showed no difference between consistent and inconsistent critical words, suggesting that they simultaneously compared information against both possible worlds. These results support previous dualrepresentation accounts for counterfactuals, and provide new evidence that linguistic cues can guide the reader in selecting which world model to evaluate incoming information against.Crucially, we reveal evidence that maintaining and updating a hypothetical model over time relies upon the availability of cognitive resources.

Conditionals on crutches: Expanding the modal horizon

“…This pattern can be seen as supporting the notion that the interpretation of counterfactuals is strict, as assumed by Lewis (1973), and that most counterfactuals are false Hájek (2014): It is difficult to prove and easy to disprove them, especially when they are based on outlandish premises. We have also found evidence that implicates working memory in counterfactual processing, which is expected if mental simulation of possible worlds is involved, and matches previous evidence from the processing literature (Ferguson and Cane, 2015). Based on the finding that high-capacity participants are more likely to give FALSE judgments for counterfactuals, and especially in implausible cases, our preliminary conclusion is that these individuals may strategically allocate mental resources to falsification, which increases their efficiency at performing the task.…”

“…Readers with higher working memory capacity may engage with counterfactual statements more deeply, considering more possibilities, than readers with lower working memory capacity. Furthermore, given that mental simulation taxes working memory (Ferguson and Cane, 2015;Van Hoeck et al, 2015), high-capacity readers may be able to represent a higher number of possible worlds within their modal horizon compared to low-capacity readers. Assuming that deviation from the ostensible truth value of a given counterfactual is a signal that the modal horizon has been expanded, high-capacity readers should thus show more deviations than low-capacity readers.…”

Understanding Counterfactuality: A Review of Experimental Evidence for the Dual Meaning of Counterfactuals

“…Subjects' sensitivity to contextual constraints and other pragmatic abilities (e.g. Ferguson et al 2014;Nieuwland et al 2010) as well as working memory capacity (Ferguson and Cane 2015) are possible candidates.…”

Plausibility and Perspective Influence the Processing of Counterfactual Narratives

“…Our results show that within the short counterfactual sentences used here, readers favoured the counterfactual world for processing incoming information online. However, given the large body of empirical evidence that has demonstrated dual representations for counterfactuals (Ferguson, 2012;Ferguson & Cane, 2015de Vega et al, 2007;Gomez-Veiga et al, 2010, Santamaria et al, 2005Urrutia et al, 2012), it is expected that this initial counterfactual interpretation would be weakened if a longer delay was introduced between the counterfactual antecedent and the target word, resulting in reduced or delayed anomaly/implausibility detection effects. Further research is necessary to explore this.…”

“…That is, reading times on the critical word (first-pass fixation duration and total reading time) did not differ between counterfactual-consistent (ÒcarrotsÓ) and -inconsistent (ÒfishÓ) continuations. Further evidence for dual-representations activated by counterfactuals can be seen in Ferguson (2012), Ferguson and Cane (2015), de Vega, Urrutia, and Riffo (2007), de Vega & Urrutia, (2012), Gomez-Veiga, Garcia-Madruga, and Moreno-Rios (2010), Santamaria, Espino, & Byrne (2005), and Urrutia, de Vega, and Bastiaansen (2012).…”