Steffen Künn used a setting that scores its own decisions. Official chess tournaments monitor the air in the playing hall and every move can be scored objectively against an engine. Across 30,000 moves, each 10 micrograms per cubic metre more fine particulate matter indoors raised the probability of an erroneous move by 26.3 per cent. Breaking the games down by stage showed the damage growing under time pressure, and the finding replicated on a national league dataset.
Armin Jeddi Yeganeh screened 45 laboratory reports published between 1980 and 2018 and pooled the 28 that met the criteria. Against the control temperature, a weighted mean rise of 4.34 °C took about 0.40 per cent off cognitive performance, 10.04 °C took 5.37 per cent and 26.68 °C took 7.97. The curve is bell-shaped around the control temperature, so cold does it too. Attention-demanding tasks suffered most, and accuracy suffered more than speed as exposures lengthened. These are laboratory ranges: a rise of 26.68 °C is not a temperature any office reaches, which is part of why the review on the previous sheet finds nothing across the narrow band an office lives in.
Kristin Yeoman put eight people through treadmill exercise and then four cognitive tasks in a chamber, first at 21.1 °C and then twice at 37.8. Reaction time went from 343.7 milliseconds in the control to 366.3 and 379.0 in the two hot trials. Counting only the responses slower than 500 milliseconds, the odds of a lapse were 1.44 times the control in the first hot trial and 3.13 times in the second. Mean lapses per ten-minute test were 2.14, 2.7 and 5.5. Eight people is a feasibility study, and the paper presents it as one.
Yuyan Chen put 32 people on a treadmill in a chamber at 26, 30, 33 and 37 °C with 70 per cent humidity and ran neurobehavioural tests for 90 minutes. At 37 °C accuracy on semantic interference and visual perception fell significantly after 45 minutes, while speed rose, which is the pattern of someone hurrying and getting it wrong. The changes tracked tympanic temperature, heart rate and dehydration. Chen turns it into a recommendation: 45 minutes as the safe span for continuous moderate-intensity work at that temperature.
Dylan Jones ran five experiments in 1990 on whether loudness or meaning does the damage. Irrelevant speech impaired proofreading only when it was meaningful, and only when the load on short-term memory was low. Neither loudness between 50 and 70 dB(A) nor the spatial position of the sound made any difference at all. A sixth experiment found serial recall of visual lists impaired by any speech-like sound, including reversed speech and speech in an unknown language, which Jones treats as a different phenomenon from the proofreading one.
Künn’s 30,000 moves are the largest observation here, and the only one from a real setting with a real stake. Jeddi Yeganeh pools 28 laboratory studies. Chen worked with 32 people and Yeoman with eight. Jones reports no participant numbers in the abstract available here.