|
|
TITLE
INTRODUCTION
SUBJECTS AND METHODS
RECORDING
STATISTICAL ANALYSIS
BEHAVIOUR AND PERFORMANCE REULTS
DECISION TIME RESULTS
EVENT RELATED POTENTIALS RESULTS
DISCUSSION
CONCLUSION
ADDRESS
TITLE:
COGNITIVE EVENT RELATED POTENTIALS DURING A LEARNING TASK M. F. EL-BAB and E. M. Sedgwick ABSTRACT: Event related potentials were recorded from 25 scalp electrodes on 44 normal subjects (Age 21-34 yr.). Our aim was to determine whether there was a difference in potentials in those that learned (successful learners), compared to those who did not (unsuccessful), and a control group (Observers). The word cognitive comes from the Latin word "cognoscere", meaning to know and this indicates what cognitive psychophysiology is all about. 34 adult volunteers aged 21:34 years (15 females), performed the test. Ten subjects passively observed the screen (control group). Subjects were seated comfort in front of a computer monitor (IBM 14 inch) in isolated room, temperature controlled. Typical learning trial proceeded as follows FIGURE 1 : For each session the subject was presented with 200 random images of two types (A or B). Image exposure was for 2000 msec followed by 500 msec blank grey screen. 25 scalp recording sites include all the standard international 10-20 system location
FIGURE 7A , FIGURE 7B , FIGURE 7C , and FIGURE 7D . The impedance between each recording site and Linked mastoid reference was reduced to below 5Kohms. Recording bandwidth was 0.03 - 30 HZ.
Sweeps with eye artefact were rejected. Digitising was at 250/sec for 625 points. We have grand averaged the all, correct, incorrect, first and last 50 ERP sweeps and decision time (DT) in the learners and the non-learners group. We have grand averaged all, first fifty and last fifty ERPs sweeps in the observers (control) group. The first time window 250 msecs to 550 msecs and the second time window 550 msecs to 850 msecs. The ERPs were measured, quantified & compared by calculating the mean amplitudes during the two time windows. The mean amplitudes were compared by Independent samples & paired T-test and repeated measures ANOVA-models. Latency and amplitude increase and decrease were studied and assessed by inspection through the event related potentials waves morphology (positive and negative deflections), Comparison was made between the grand averaged all sweeps, correct & incorrect responses sweeps, first & last 50 sweeps, and right and left side sweeps too. The performance of the subject is seldom perfect; it is usual to allow a certain tolerance 10% (0.1) failure rate (red line ). A subject performance with 90% success will generate a horizontal and the line will move above horizontal for a worse performance. Individual increments for the cusum would then be 1.0 - 0.1 = 0.9 for each failure, and 0 - 0.1 = -0.1 for each success, and the same equation for tolerance 20 % (blue line), and tolerance 30% (pink line). The Cusum statistical method has been used to assess the subject's performance. The cut off for learning was set at 3 70% correct for the last fifty answers, and it's profile was divided into two categories, the first one subject reflecting satisfactory progress or better (learners, n=18) and the second one subject considered unsatisfactory (non-learners, n=16). The Cusum charts by visual inspection showed that all subjects have relatively poor performance at the beginning of the trials. For the learners they had effective after the first fifty FIGURE 8 . A typical cusum chart from non-learner who struggles to achieve the required success FIGURE 9 . In the learning group decision time significantly decreased during learning, but increased for Non-learners FIGURE 14. We Grand averaged the learners (n = 18) and the non-learners (n = 16) all, first fifty, last fifty, correct, and incorrect answers sweep. We found that the prestimulus baselines of the waveforms are very similar, and the sensory portions of the waveforms are similar too Observers showed less positivity than those trying to learn
FIGURE 15 . The ERP waves begin to change at 250 msecs and show marked differences from 300 msecs to 800 msecs. There is increase in the positivity occurring from 300 msecs onwards in all trial displays in the learning group, and there is less positivity for the group of non-learners. In the case of the observers group, the ERPs waves appear to be close to the baseline throughout the test FIGURE 16 . There are highly significant differences of the P300 waveform amplitude between the learners and non-learners, and between both of these two groups and the observers group during the two time windows. We are comparing the first 50 of the 200 images with the last fifty of these images. In the first fifty we expect they are guessing, as they have not had the chance to learn yet and the last 50 if we can assume that they either have or have not learned. If the subject has learnt to distinguish the two stimuli you would suspect the neurophysiological processes from the start to end answers waves and correct to incorrect answers waves to be different. In the group of learners start to end (before and after learning) there does appear to be different with increase in the positivity occurring from 200 msecs onwards in the last fifty displays FIGURE 17 , and does not for the non-learners FIGURE 18 . The last fifty ERP's differences can be seen in FIGURE 19 , A deviation in cognitive evoked potential In the learners correct to incorrect answers, can be seen from 200 msecs to 900 msecs FIGURE 20 . Whereas in the case of non-learners group, the activity appears to be similar throughout the test FIGURE 21 .
FIGURE 22 ,shows the correct answers differences between the learners group and the non-learners group during the two time windows. The complementary pattern used as stimuli, although forming two categories, was very similar visually. This sensory VEP and the cognitive portion of the ERP waves show that processes evoked by either pattern are similar in the both groups. Learners FIGURE 23 .and non-learners FIGURE 24 . Right side and left side in the learning group FIGURE 25 , in the non-learning groupFIGURE 26 , A brainmap shows the frontal medline differences between learners FIGURE 27 , and non-learners. FIGURE 28 . Subtraction between both of the learners group and the non-learners group minus the observers group the two time windows brainmaps FIGURE 29 shows the first time window brainmap FIGURE 30 shows the second time window brainmap,and the ERP's wave subtractions, for learning group minus the all ERP waves for the non-learning group and the all ERP waves for the learners and non-learners minus the all ERP waves for the observers group FIGURE 31 ERP ,
The mean amplitude in the learners, the non-learners, and the observers groups during the two time windows FIGURE 32 shows the first time wimdow, and FIGURE 33 shows the second time window. This increased P300 during learning correlate with our hypothesis that there is a difference in brain activity as a result of learning. Our aims of this study, to look for the change in cognitive event related potentials, at the beginning and at the end, and at the correct & incorrect answers, of learning task in which subjects learnt and did not, to examine the decision time, and subjects performance. There is no literature that correlates all of these data in one set of participants. PERFORMANCE The ideal subjects seemed to be a moderately motivated and competitive individual. During de-briefing, motivation and function seemed to contribute to performance, certain individuals appear to be over competitive or to try too hard resulting incorrect answers provoking frustration. The non-learner group seemed under motivated and they didn't want to be there, were thinking of other things and couldn't focus on the task. Learning process appeared to go through three phases: Subjects were guessing, resulting in 50% performance and string of three or four similar answer at most. Subjects' tactics were to compare and distinguish between the past image and the present pattern on the screen. Learners constructed a more robust model and have associated the patterns to an individual mouse button depend on the previous answers. Non-learners either stayed in stage 2 or revert to stage 1 due to frustration or surrender. Subjects employed strategies and beliefs and the classification task was at a subconscious level. Only two out of eighteen learners were close in their explanations of classification of patterns. DECISION TIME The decision time was quicker for the learners than non-learners. The reduction in the decision times of the learners towards the end of the trials becomes shorter as well as more accurate, suggested that knowing the answers and responding is quicker process than guessing answers. For the correct/ and incorrect answers decision time were significantly different, for the learners not for the non-learners. That finding is in line with the findings of (Fernandez, et al 1998), that the reaction time significantly different between the correct and incorrect answers for the group with good performance, Thus learning results from tactic employed within this extra expose time of the stimuli. The fact that the non-learners showed on improvement or correct/incorrect variance reinforced the hypothesis that, learning mechanisms effected, this increased processing speed. Corresponding with a start, and end ERPs difference, learners showed a significantly shorter decision time to non-learners at the beginning, and the end of the trial. There were no significant differences between the subjects decision time either in the learners group or the non-learners group for type A and Type B patterns answers, which result in, that there is no difference between these images structures and task hardness, and that what confirmed by the ERPs waves. In the learners and non-learners decision time for type A & type B there are highly significant differences result in the differences in both group strategies and tactics. EVENT RELATED POTENTIALS Generally the ERP Results reflect several findings related to the impact of learning on the underlying brain circuitry. The first and second time windows the observers did not show any positive activity. That changes in ERPs most probably related to the learning processes, and the decision-making processes in the learners and non-learners group, which is not applicable to the observers group who did not ask to learn or to make any decision. First time window P300 is most positive in the learning group at the end of the trial, it has been associated with increase memory effects, depth of processing and comparison of stimuli. P300 has been associated with increased memory effect, depth of processing and comparisons of stimuli. The beginning of the trial for learners shows differences ERP's of non-learners. The unique feature of this ERP is the negative peak around 700 msecs; the learner's waves more negatives than the non-learners waves. This initial negativity could be hypothesised to be the result of a process that is essential for later learning. For example a foundation block for a robust model of learning. This would insinuate that a prediction could be made from an ERP at the beginning of the trial as to whether a subject will learn, possibly reflecting a difference in tactic involved. After learning or for the last fifty answers waves in the learners group. P300 amplitude was larger for that group more than the non-learners, starting around 250 msecs were. Another late positive peak around 500 msecs, which give us impression that the learning processes, represents by that later positivity. Reaction time quicker after learning a categorical perception task compared with before learning and in non-learning subjects. Successful learning is associated with an increase in late positivity. No comparable changes are seen amongst non-learner who have undertaken the same task. The psychological processes involved in learning can be measured by using ERPs. This technique with task modifications could be used in the investigation of patients with learning difficulties.
Recording were from cap electrodes with linked mastoids as reference, bandwidth 0.03 - 30.00 HZ.
In two hundred different trials obeyed the rule for patterns A and B which had been generated and presented randomly by computer program and classed according to the subject's decision by pressing buttons.
Clues from border effect or contrast change were eliminated. The subject's performance was assessed by the CUSUM statistical methods.
18 learned better than 70% correct judge from the last 50 trials did. Ten other subjects passively observed the patterns.
The total, the first & the last 50, the correct & the incorrect wave responses mean amplitudes and mean Decision times were measured,averaged and compared.
The observers group ERPs was very similar to the base line activity. Learners showed an increase in positivity, whereas non-learners did not show a significant increase in positivity during any of the windows.
The increased positivity was greater for correct answers than for wrong answers and more marked frontally and on the right hemisphere than the left.
The learning process is associated with scalp positivity which is greater in successful compared with unsuccessful subjects.
Also the categorical learning is predominantly a right hemisphere activity. Reaction time was quicker for the learners.
The kind of evoked potential is the sequences of voltage changes generated in the brain, and in the sense organs and pathway leading to the brain, following the reception of the transient physical stimulus.
Learning and memory are different sides of the same coin. Memories are what left behind as a result of learning, and we infer the existence of learning from the presence of memories.
The aim was to develop a learning task in which some people learnt and some people didn't, ERPs were to be recorded before and after learning and these average compared inter and intra-group, and the cognitive area and especially the late positive waves were our interest.
Each image is made from a random arrangement of 40 x 32 letter-like elements of four different forms. Type A or B images contain 12.5 % more or less of one of the elements. There was no overall luminance change. FIGURE2 , and FIGURE3 Subject made a single response to each image as soon as possible within the image display time followed by right feedback FIGURE4 ,or wrong feedback FIGURE5 , on the screen. The same images number (200), and patterns (A & B) were displayed for the control group. Were just asked to watch the screen without either decision making or pressing buttons FIGURE 6 .
ERPs recording procedures FIGURE 7E .
The CUSUM ststistical method used to assess the subjects performance. The CuSum charts starts at zero, declining in the cusum trend indicate success, and an increasing in the cusum trend indicate failure. We have derived learning curves from the results of the random visual stimulus experiment, by assigning the value 0 to each success, and the value 1.0 to each failure.
This may be express mathematically as:-
Designations of position should be in terms of brain areas (Frontal, Parietal, Temporal, and Occipital) rather than only in numbers so the communication will be more meaningful to the non-speacialist.
There were some subjects in non-learning group seem that they had got in the middle of the trial the difference and they were able to identify between both patterns (A&B), but they lost that again FIGURE 10 , and another subject character in the learning group who was not able to identify the difference between both patterns (A&B), but by the end of the trials he got the clue FIGURE 11 .
The subjects performed three responses correct, incorrect, and no response. A clear divide were seen. Two learners performance only were > 70%, three non-learners were > 50%, and non of them was > 60% FIGURE 12 . Learners performance shows steady increase, and as expected the non-learners exhibited no start to finish difference (Learning curve FIGURE 13 ). For each cusum Chart Y-axis represents the performance, and X-axis represents the 200 trials.
Decision time was longer for incorrect answers, and first fifty answers in both groups. The decision time was similar in the learners group for type A and type B in every group, and it is shorter for type B in the non learners.
The first time window 250 msecs to 550 msecs and the second time window 550 msecs to 850 msecs. The ERP graphs show a timeline of the average brain activity in all electrodes. The Y-axis is measuring the activity in microvolts; the X-axis is time scale from -500 to 2000 msecs.
The correct answer waves in the learners group were more positive at 300msecs and 500 msecs than the incorrect answers in the learners, and the correct answers in the non-learners. result in that the more correct answers the more early positive peaks, and due to participant or learners early and correct decisions matching with the image on the screen.
Subtraction brainmaps for these ERPs show over all voltage changes starting from 300 msecs to 600 msecs. Which are related to the learning processes. For learner, brainmaps of subtraction for start and finish ERP show a major frontal, parietal, and temporal positivity that predominantly right hand side. The subtraction for the second time window is more positive due to cumulative changes. The changes from start to finish ERP and the voltages represented by subtraction brain maps are representations of learning. No other variable can account for these results. Non-learners have been through the same task as the learners but don't show the ERP changes. There additional positivity in the ERPs waves after the feedback which given to the subjects by them response specially in the learners group for correct answers. Result in using the information given to the subjects by the feedback to modify them coming responses, and it is more positive in the learning group due to they expecting that feedback which are most correct most of the answers. From data comparing the hemispheres a major difference is seen. The right hemisphere, in conjunction with other visual and categorical perception tasks, shows increased positivity from 300 msecs onwards. The P300 event related potentials (ERPs) is considered to be closely related to cognitive processes, P300 scalp latencies increase, parietal P300 scalp amplitude decrease and the scalp potential field shift to relatively more frontal distribution.
The learning task set is difficult and only about half of the subject learned. Even so, they could not express verbally what they had learned. The P300 was increased after learning, further experiments are needed to see if P300 amplitude decreases when the task has been already learned.
HOME
