The Neural Mechanism Of Antonym Superiority Effect

The “Antonym Superiority Effect” has attracted great attention over the past decades.Comparative studies of synonyms and antonyms have found a significantly faster reaction to antonyms than to synonyms,while the neuroimaging results have also demonstrated different neural mechanism of antonyms and synonyms.However,there remains considerable controversy over the neural mechanism of its processing.The focus of the controversy is: does the polarity feature or the lexical association of antonyms cause the superiority effect? The“Semantic Relation Hypothesis” proposed by Kadesh et al.holds that the “Antonym Superiority Effect” results from its semantic opposition and polarity.However,according to the “Lexical Association Hypothesis” proposed by Miller et al.,the “Antonym Superiority Effect” depends on the lexical association as canonical antonyms co-occur more frequently than canonical synonyms,which results in antonyms’ more solid entrenchment in semantic memory and stronger lexical association.The controversy has not been resolved because semantic relation always intertwines with lexical association.Adopting Chinese compounds and non-words formed by antonymous/synonymous characters,the present study dissociates semantic relation and lexical association to further investigate the “Antonym Superiority Effect” in visual word recognition.The present research is primarily concerned with the following two questions:（1）Do lexical associations（strong/weak）influence the visual word recognition?（2）Do semantic relations（antonym/synonym）influence the visual word recognition? Through exploring the answers to the above two questions,the present study further discussed the applicability of the “Semantic Relation Hypothesis”,“Lexical Association Hypothesis” and the “Associative Boost Hypothesis”.Semantic relations（antonymy and synonymy）and lexical associations（strong association and weak association）were selected as independent variables,forming a 2×2 design.There were four conditions,i.e.,antonymous strong association condition（e.g.,qi3-luo4,rise-fall,the compound means “rise and fall”）,antonymous weak association condition（e.g.,ran2-xi1,burnextinguish）,synonymous strong association condition（e.g.,ben1-pao3,rush-run,the compound means “run”）and synonymous weak association condition（e.g.,gu4-pin4,hireemploy）.The results of Experiment 1 found that（1）the main effect of semantic relation（synonymy/antonymy）was not significant while the main effect of lexical association（strong/weak association）was significant,and the reaction to the strong lexical association condition was significantly faster than that of the weak one.The interactive effect of semantic relation and lexical association was significant as well;and（2）The simple effect analysis showed that the reaction to the strong association condition was faster than the weak one for the antonym condition,but there was no significance between the strong and weak association conditions for the synonym condition.Besides,the reaction time of the strong association condition was shorter than the weak one for the antonym condition,but there was no significance between the strong and weak association conditions for the synonym condition.The results of Experiment 2 were as follows:（1）The semantic relation mainly played an important role in the Late-N400 time window and affected word processing until the P600 time window.Furthermore,under the strong association condition,the “Antonym Superiority Effect”appeared in the Late-N400 and P600 time windows.But under the weak association condition,only in the Late-N400 amplitude did the “Antonym Superiority Effect” appear.（2）The lexical association affected the whole process of word recognition and processing.Furthermore,under the antonym condition,the “Strong Association Effect” appeared in the P200,Early-N400,Late-N400 and P600 time windows.But under the synonym condition,only in the Early-N400 and Late-N400 time windows did the “Strong Association Effect” appear.The conclusions are as follows.Firstly,the present study does not support the “Semantic Relation Hypothesis” proposed by Kadesh et al.because the “Antonym Superiority Effects” for the strong and weak association conditions are different.Secondly,the present study does not support the “Co-occurrence Hypothesis” proposed by Miller et al.either because the “Strong Association Effects” for the antonym and synonym conditions are different.Finally,the present study supports the “Associative Boost Model” proposed by Moss et al.as the “Antonym Superiority Effect” is greater for the strong association condition and the “Strong Association Effect” is greater for the antonym condition.