Manipulability in disguise: Uncovering the behavioral and neural differences in processing words representing small and big objects

Size is a fundamental visual-spatial characteristic of the physical world. Previous studies have shown distinctive neural responses along the ventral temporal cortex when processing pictures representing small and big objects. However, it remains unclear whether this size-based categorization is limited to perceptual information or extends to symbolic input, such as words. Furthermore, some behavioral studies have indicated that words denoting big objects (big words) were recognized faster than those denoting small objects (small words), but this effect remains controversial. Therefore, the present study investigates both the behavioral and neural differences in processing small and big words, and whether these differences are task-dependent. Behavioral results revealed a significant processing advantage for big words in the lexical decision task (LDT), evidenced by shorter reaction times and lower error rates. However, this advantage was not observed in the semantic decision task (SDT) in which participants were asked to judge whether a word was a concrete word. The multivariate pattern and time generalization analyses of EEG data showed distinguishable and stable neural patterns for processing small and big words between 190 ms and 250 ms in both tasks. Additionally, in the SDT, these neural patterns persisted from 390 ms to 520 ms, reflecting a reactivation of earlier neural representations. Further analysis using multiple RSA regression showed that manipulability, rather than size, was the primary factor explaining the neural responses, with the significant time windows consistent with the above-chance decoding time windows in both tasks. These results suggest that the processing advantage to big words may acutally relate to manipulabiltiy, rather than object size.These findings also highlight a flexible and two-stage processing of semantic concepts, with shared neural representations between these two stages.

Experimental procedure

Behavioral results

MVPA and time generalization results

SVM was applied to classify the nerual patterns of processing small and big words

RSA and factor analysis

Multiple RSA regression results

Experiment 1 and 2 were behavioral experiments in which the lexical decision (LD) and semantic decision (SD) tasks were carried out separately, each with 60 participants. Experiment 3 and 4 were the corresponding EEG experiments using the same tasks, each with 26 participants. Both by-participant and by-item paited t-tests were used.