"Perturbation of Old Knowledge Precedes Integration of New Knowledge"

 Xiaoping Fang

The importance of consolidation in integrating new knowledge has received much recent attention in the field of word learning. Less examined is the change of existing word knowledge as a result of learning, which possibly occurs even prior to the opportunity for consolidation. In the current study, participants learned new meanings for known words and novel words. Then they performed a one-back task on a list of words among which the trained words were followed by probes of new and original meanings, while EEG was recorded. Consistent with one account of complementary learning systems, familiar words related to the new meanings of preceding trained words did not show a N400 reduction compared to unrelated words. However, when preceding known words had been paired with new meanings, in contrast to being presented as exposure controls, a larger negativity (especially between 550 and 750 ms) was found for words related to the original meanings. Together, the results suggest that even before a new meaning has become integrated with a word form, meaning learning perturbs the word form’s connection with the original meaning.

"Stimulus Onset Asynchrony (SOA) affects Error-Related Negativity (ERN) Morphology to a Semantic Task"

 Erika Laing

 While Error-Related Negativities (ERNs) are traditionally elicited by sensory-motor stimuli, it has recently shown that semantic stimuli evoke an ERN when embedded in a speeded semantic category judgment task (Balass et al, 2015; Laing et al, in prep).  Unlike sensory-motor tasks, semantic category judgments additionally engage language and memory probing processes, which increase the complexity of decision-making.  In this more complicated task space, it is unknown how variations of allowable response time affect the response-related neural activity.  Two studies were performed to assess in what way varying the time pressure may affect the morphology of the Error-Related Negativity.  Both studies used an identical semantic category judgment task wherein the participant simply responded “yes” or “no” depending on if each presented word belonged to a category defined at the beginning of each block.  Unbeknownst to the participant, category member words were varied in their typicality, and foils were varied in their semantic relatedness.  For example, in a category FISH, a Typical Category Member might be trout, Atypical piranha, Related Foil gill, and Unrelated Foil acrobat.  Study 1 was a highly time pressured task, with just 600ms within which to see the word and respond (350ms on screen, followed by 250ms blank), while Study 2 was somewhat more relaxed with 800ms total (350ms word, 450ms blank).  Both studies showed ERN’s to Typical Category Members (e.g. – trout) and Unrelated Foils (e.g. – acrobat), but not to Atypical Category Members or Related Foils.  These results support a multi-stage category membership validation process that predicts that typicality and unrelated foil decisions carry strong signals of category membership status, while atypical category members and related foils require additional processing to validate their status.  Critically, this additional processing time is interrupted by the speeded task state, leading to the absence of a differentiable error signal reflected by the ERN.  The multi-stage validation process theory is further supported by the subsequent Response Positivity (Pe) elicited by correct responses, wherein Typical Category Members and Unrelated Foils are more highly positive than Atypical Category Members and Related Foils.  While these patterns of results were similar across the two studies, the latency of the ERN was substantially different between the two: longer SOAs led to more quickly elicited ERNs (Study 1 ERN: 68ms, Study 2: 28ms).  Latency of the response-related activities therefore may reflect the relative amount of pressure that task environment exerts on the error