Top-down feedback normalizes distortion in early visual word recognition: Insights from masked priming

Area of Science:

• Cognitive Psychology
• Neuroscience
• Computational Linguistics

Background:

• Visual word recognition exhibits tolerance to input distortions, as shown by priming effects with CAPTCHA-like stimuli.
• This tolerance is explained by either bottom-up normalization or top-down lexical feedback mechanisms.
• Previous research has not fully disentangled the contributions of these two theoretical accounts.

Purpose of the Study:

• To investigate the role of top-down lexical feedback in visual word recognition when presented with distorted stimuli.
• To differentiate between bottom-up and top-down processing models in early stages of word recognition.
• To examine the influence of word frequency on the tolerance to visual distortions.

Main Methods:

• Conducted two masked identity-priming experiments using both printed and CAPTCHA-like distorted word primes.
• Utilized a lexical decision task and a same-different matching task to assess priming effects.
• Manipulated word frequency (high vs. low) to test differential effects of top-down feedback.

Main Results:

• Identity-priming effects for high-frequency words showed minimal difference between printed and distorted primes.
• A larger difference in priming effects between printed and distorted primes was observed for low-frequency words.
• In a same-different matching task (bypassing lexical access), priming was greater for printed primes and unaffected by word frequency.

Conclusions:

• Top-down lexical feedback appears to play a crucial role in normalizing visual input during lexical access.
• The findings challenge purely bottom-up models of visual word recognition, suggesting an interaction between feedback and feedforward processes.
• Word frequency modulates the reliance on top-down feedback for processing distorted visual word stimuli.