Since the invention of airconditioning over 100 years ago a central research challenge has been to define the indoor environmental temperatures best suited for occupants. The first scientific approach to this question was framed in terms of optimising occupant thermal comfort, commonly expressed as a U-function, symmetrical around a single optimum temperature for any given combination of the remaining comfort parameters (ISO, 2005). The inescapable conclusion drawn from such logic in the minds of risk-averse design engineers is that the only strategy able to reliably deliver occupant comfort is HVAC applied to sealed-façade architecture. A rigorous scientific rebuttal of the “single temperature optimum” model of comfort came 30 years after PMV/PPD was first floated (e.g. de Dear and Brager, 1998; 2001). Known as the adaptive comfort model, a clear implication is that passive design solutions are capable of delivering comfortable internal environments across a broad swathe of climate zones, throughout most if not all of the year. But recently the “single temperature optimum” model has resurfaced, this time with its justification shifting away from the thermal comfort requirements of occupants towards their cognitive performance. Beyond the building science domain, in disciplines such as psychology and ergonomics, the prevailing wisdom regarding temperature effects on cognitive performance is an extended-U rather than an inverted U function. The gist of the model is that cognitive performance is relatively stable throughout the moderate temperature range, but it rapidly deteriorates at the boundaries of thermal acceptability where stress drains the performers’ attentional resources. The extended-U model has garnered broad acceptance across a range of disciplines with the notable exception of HVAC engineering and indoor air sciences. But the weight of research evidence tends to support the extended- rather than inverted-U model. In this paper the arguments regarding thermal effects on cognitive performance are critically evaluated.