Human Delayed Matching-to-Sample Performance Under Normoxic and Hypoxic Conditions

Abstract

The impact of mild to moderate levels of hypoxia (reduced blood oxygen) on cognitive function is highly controversial. The literature is laden with contradictory findings on the degree of hypoxia necessary to show an effect, the size of the effect, and even the direction of the effect. Hypoxia is a significant factor in many common situations (such as flying, climbing) and in many medical conditions. Given this, it is imperative that the cognitive response to hypoxia is understood so that appropriate safety guidelines can be put in place. The present research applied behavioural detection theory (Davison & Tustin, 1978) to the study of hypoxia. In three experiments, human delayed matching-to-sample (DMTS) performance was investigated under normoxic and hypoxic conditions. Experiment t had six subjects complete five 50 minute sessions on a DMTS task requiring visual discrimination of various line-lengths, under normoxic conditions. Forgetting functions (discriminability as a function of retention interval) were described well by several models, especially a time-scaled negative exponential model. This model allowed estimation of initial discriminability between the sample line-lengths with no retention interval (log d0) and the rate of decrement in discriminability with increasing retention interval (a). Log d0 increased with experience on the task, however there were no systematic changes in a. The time taken to choose the comparison stimulus increased with retention interval, but did not change systematically over the five sessions. Confidence in response decisions decreased with retention interval and increased with experience on the task. Response bias (log b) did not change with retention interval, however it changed systematically over the five sessions, from a bias towards reporting short line-lengths to a bias toward reporting long line-lengths. Experiment 2 took the DMTS task used in Experiment 1 and tested six subjects at three levels of normoxia / hypoxia. After training, subjects completed two experimental sessions breathing gases equivalent to sea level, two sessions breathing gases which brought their blood oxygen saturation (SpO2) to the mean level experienced at 10000 feet (3048 metres) in altitude, and two sessions breathing gases which brought their SpO2 to the mean level experienced at 14000 feet (4267 metres) in altitude. While many of the normoxie findings from Experiment 1 were replicated, there were only two consistent effects of hypoxia across subjects: heart rate increased, and subjects were more confident. Experiment 3 improved the DMTS design used in Experiments I and 2 by controlling the interval between consecutive trials. Six subjects, experienced with the DMTS task, completed a "No Interference" condition and an "Interference" condition where a tracking task was interpolated into the retention interval. Effects of controlling the intertrial interval, proactive interference from earlier trials, and retroactive interference from the tracking task were assessed. Each condition was completed at the same levels of normoxia / hypoxia tested in Experiment2. Hypoxia did not have a reliable effect on confidence or heart rate in this experiment. However, it had three main consistent effects: Response times slowed (in the "No Interference" condition); tracking performance deteriorated; and respiration rate slowed and became more variable. Many of the normoxic findings in Experiment 1 were replicated, and there were several additional findings. First, controlling the intertrial interval did not have a consistent effect on performance. Secondly, the tracking task did not retroactively interfere with performance, but led to higher log d0 estimates and faster response times. Thirdly, proactive interference from the sample presented on trial N-l led to an increase in a in trial N when data from the two conditions were combined, but no change in log d0. Finally, subjects became more biased toward reporting the long comparison line-length with increasing retention interval in the "No interference" condition, but not in the "Interference" condition. Behavioural detection theory and the time-scaled negative exponential function proved to be very useful analytical tools for gaining insight into human memory performance, under both normoxic and hypoxic conditions. Individual responses to hypoxia were generally quite different, with very few group effects obtained. Further research exploring physiological and psychological factors which could help explain these individual differences is required.