Temporal Shift Length and Antecedent Occurrence Likelihood Modulate Counterfactual Conditional Comprehension: Evidence from Event-Related Potentials

Abstract

Counterfactual conditionals posit hypothetical scenarios in which antecedent events contradict reality. This study examined whether and how the processing difficulty of Chinese counterfactual conditionals (yaobushi, equivalent to if it had not been for in English) can be affected by the length of temporal shifts of the events across clauses and the likelihood of the antecedent occurrence. Participants read Chinese counterfactuals that contained either long (e.g., qunian-xianzai [last year-right now]) or short temporal shifts (e.g., zuotian-xianzai [yesterday-right now]) within highly likely (e.g., sign up for school activity) or less likely contexts (e.g., sign up for Arctic scientific research). ERP results revealed a significant N400 interaction between the temporal shift length and antecedent likelihood on the temporal indicators in the consequent and the sentence-ending verbs. Specifically, the less likely events elicited larger negativity than highly likely events with short temporal shifts on the temporal indicator. On the sentence-ending word, the long temporal shift elicited enlarged negativity than the short temporal shift when the antecedent was highly likely. These findings have two key implications regarding the interplay of implied causality and falsity constraints during counterfactual comprehension. First, salient falsity constraints can override effects of causal coherence on processing. Second, greater negativity for unlikely antecedents suggests that counterfactual markers concurrently activate factual and hypothetical representations.

Keywords: Chinese counterfactuals; N400; implied causal relationship; likelihood of occurrence; temporal shift.

Figure 1

Stimulus presentation sequence in the present experiment.

Figure 2

Grand average ERP waveforms time-locked to the onset of the temporal indicator of the consequent from −200 to 800 ms at a representative electrode. (a) represents the grand average waveform for all four conditions during the time window of −200 to 800 ms on a representative electrode. (b) illustrates the topographic maps of the mean amplitude difference between the “Less likely, short” and “Highly likely, short” conditions in corresponding time windows.

Figure 3

Grand average ERP waveforms time−locked to the onset of the sentence−ending word of the consequent from −200 to 800 ms at a representative electrode. (a) represents the grand average waveform for all four conditions during the time window of −200 to 800 ms on a representative electrode. (b) illustrates the topographic maps of the mean amplitude difference between the “Less likely, short” and “Highly likely, short” conditions in corresponding time windows.