The impact of string length on neural categorization processes for letters and digits

Letters and Arabic digits are the building blocks of words and numbers. In the visual cortex, these culturally acquired characters are characterized by a differential involvement of the left and right hemispheres. Letters, as language-related symbols, predominantly involve left-hemispheric structures in the occipito-temporal cortex, while digits, as quantity-related symbols, elicit right-hemispheric or bilateral visual recognition processes. However, it remains unclear whether the human brain processes single elements and strings of characters differently depending on their category. This question is important because in the Latin alphabet, letters are usually combined in strings to form words and do not stand alone, while digits have meaning in both cases. Using Fast Periodic Visual Stimulation (frequency-tagging) during EEG recordings, we investigated how adults (N = 18) discriminate letters and digits from each other, as a function of their string length (i.e. 1 vs 5 characters). One category of stimuli (e.g., single letters) was periodically inserted (1/5) in a stream of stimuli of the other category (e.g., single digits) displayed at 10 Hz. Results showed clear discrimination responses at 2 Hz (i.e., 10 Hz/5) with occipito-temporal topography, stronger for strings than for single elements. Digits gave rise to right-lateralized responses whatever the length. Letters displayed a left-lateralized topography only when strings were presented, while single letters were right-lateralized. A second experiment (N = 20) replicated these novel and unexpected findings. The results are discussed as potentially indicating that expert readers perceive single letters as visual objects of expertise, whereas letter strings engage linguistic (orthographic, phonological) processes that rely on the left hemisphere.