Incorrect responses resulted in the computer presenting the spoken words "no, this is not it", followed by an error correction procedure (trial repetition).
A single-case adapted alternating treatments (Sindelar, Rosenberg, & Wilson, 1985) embedded in a multiple baseline across word sets design was employed to examine the effects of error correction during probe trials in order to reduce error rates.
Again, across the group there was a correlation between the asymmetry of error correction on one trial and the asymmetric shift in reaching direction on the next trial.
The fact that blocked [10], [38] and trial-by-trial feedback (our study) show similar effects on learning is further support for an interpretation that casts feedback in the role of monitoring performance rather than trial-by-trial error correction.
Like the recurrent supervised model discussed above [32], we propose feedback may be used to monitor performance rather than trial-by-trial error correction, by evaluating the expected performance in relation to the perceived difficulty of the task [33].
If listeners use internally-generated feedback for trial-by-trial error correction, but cannot ignore the conflicting external feedback, changing the amount of external feedback should modulate its correlation with the internal signal, and hence the amount of improvement observed on the task.
The experimental procedure was the same as that followed in Experiment 1 and was the same for both chimpanzee and human participants, except that humans did not receive a food reward and correction trials after error trials.
This study directly examined the effects of error correction versus no error correction during probe trials on the effectiveness and efficiency of simultaneous prompting on the acquisition of sight words by three middle school students with moderate intellectual disabilities.
Baseline stimuli were the same as those used in Experiment 2. If the participant made an error, only the target appeared in the subsequent correction trials.
This systematisation suggested that the system was running two learning processes simultaneously, one short-term based on trial-by-trial error corrections and one long-term quantified across days of training.
