“Breathing Carbon Dioxide (4% for 1 Hour) Slows Response Selection, Not Stimulus Encoding”, Max Vercruyssen2014 (CO₂)⁠:

This experiment sought to determine whether breathing carbon dioxide (CO₂), a toxic environmental stressor known to impair reaction time, slows human information processing in the stimulus encoding or response selection stage, or both, and whether this effect is influenced by time-on-task and exposure duration.

In a 2 X 2 X 2 X 4 (Gas X Degradation X Compatibility X Time-on-Task) within-participants design, 6 highly practiced (more than 10,000 trials) healthy young male participants performed a serial choice reaction time (SCRT) task while breathing either 4% CO₂ (with 50% O₂) or room air (0.03% CO₂ and 21% O₂) for 60 minutes. Task variables manipulated were stimulus degradation (intact vs. degraded) and stimulus-response compatibility (high vs. low). Data from each 20-min SCRT test were subdivided into 5-min and 2-min intervals to determine the effects of time-on-task (and exposure duration).

There were statistically-significant increases in SCRT from breathing carbon dioxide (p =.004), degrading the stimulus (p <.001), lowering compatibility (p =.004), and increasing time-on-task (p =.020). Lowering compatibility served to exaggerate the impairment produced by carbon dioxide inhalation (p =.038). Time-on-task (and exposure duration), however, did not interact with gas, degradation, or compatibility.

Thus, SCRT, analyzed according to the Additive Factors Method (Sternberg1969, 1998), (1) was sensitive to the degrading effects of breathing CO₂ at an undetectable concentration that did not produce clinical symptoms, (2) determined the locus of this effect was associated with the response selection stage of processing, (3) demonstrated that the progressive deterioration in performance due to increases in time-on-task (and exposure duration) affects both the stimulus encoding and response selection stages in a similar manner, and (4) ruled out alternative explanations by showing the results did not vary with distribution analyses, data trimming, error analyses, and analyses for tradeoffs of speed-accuracy, SCRT-DT, and other performance operating characteristics.