Artificial intelligence for the measurement of vocal stereotypy

Both researchers and practitioners often rely on direct observation to measure and monitor behavior. When these behaviors are too complex or numerous to be measured in vivo, relying on direct observation using human observers increases the amount of resources required to conduct research and to monitor the effects of interventions in practice. To address this issue, we conducted a proof of concept examining whether artificial intelligence could measure vocal stereotypy in individuals with autism. More specifically, we used an artificial neural network with over 1,500 minutes of audio data from 8 different individuals to train and test models to measure vocal stereotypy. Our results showed that the artificial neural network performed adequately (i.e., session-by-session correlation near or above .80 with a human observer) in measuring engagement in vocal stereotypy for 6 of 8 participants. Additional research is needed to further improve the generalizability of the approach.     

Do goal cue and motor activity impact preschoolers’ working memory?

Preschoolers are well known for their poor working memory (WM) performance. This could result from goal neglect, which would hamper the setting of maintenance strategies. Previous studies have shown that preschoolers’ WM performance can be improved in game-like tasks, because they provide cues to support goal maintenance. However, in these studies, it was unclear what features of the task (either the main toy or the motor activity required by the game) provide efficient cues. The aim of the present study was to disentangle the two features to examine cue effects in 5- to 7-year-old children. No improvement of WM performance was observed when the toy was a potential goal cue, whereas the motor activity had a detrimental effect in all age groups. The latter effect could result from a distraction of attention from attention-based maintenance activities. Hence, preschoolers' poor WM performance would not be fundamentally due to goal neglect.     

Continuous Response Monitoring of Relative Time-to-Contact Judgments: Does Effective Information Change During an Approach Event?

In many previous studies of time-to-contact (TTC) judgments of approaching objects, a response was measured after observers viewed the entire event and was used to infer the informational basis for the judgment. Such measures primarily reflect the information used at the end of the approach event and may not reveal whether observers used different information at different times during the event. Evidence indicates that observers use multiple information sources and that the effectiveness of information varies with distance. We introduce a method in which observers continuously report which of 2 approaching objects would reach them first, throughout the approach. We identified the occurrence and time of response reversals. Most observers changed their relative TTC judgments during the event. The pattern of responses indicated that observers did not use tau early during the approach when objects were far and did not use optical size later when the objects were closer. Most observers relied on either optical size or optical expansion rate early during the approach and expansion rate later during the event, although there were individual differences. A single relative TTC judgment measured after observers view an approach event may not reflect judgments or their informational basis throughout the entire event.