Search for: All records

Thermal comfort (TC) – how comfortable or satisfied a per- son is with the temperature of her/his surroundings – is one of the key factors influencing the indoor environmental quality of schools, libraries, and offices. We conducted an experiment to explore how TC can impact students’ learning. University students (n = 25) were randomly assigned to different temperature conditions in an office environment (25◦C → 30◦C, or 30◦C → 25◦C) that were implemented using a combination of heaters and air conditioners over a 1.25 hour session. The task of the participants was to learn from tutorial videos on three different topics, and a test was given after each tutorial. The results suggest that (1) changing the room temperature by a few degrees Celsius can stat. sig. impact students’ self-reported TC; (2) the relationship between TC and learning exhibited an inverted U-curve, i.e., should be neither too uncomfortable nor too comfortable. We also explored different computer vision and sensor-based approaches to measure students’ thermal comfort automatically. We found that (3) TC can be predicted automatically either from the room temperature or from an infra-red (IR) camera of the face; however, (4) TC prediction from a normal (visible-light) web camera is highly challenging, and only limited predictive power was found in the facial expression features to predict thermal comfort. 

Automatic machine learning-based detectors of various psychological and social phenomena (e.g., emotion, stress, engagement) have great potential to advance basic science. However, when a detector d is trained to approximate an existing measurement tool (e.g., a questionnaire, observation protocol), then care must be taken when interpreting measurements collected using d since they are one step further removed from the under- lying construct. We examine how the accuracy of d, as quantified by the correlation q of d’s out- puts with the ground-truth construct U, impacts the estimated correlation between U (e.g., stress) and some other phenomenon V (e.g., academic performance). In particular: (1) We show that if the true correlation between U and V is r, then the expected sample correlation, over all vectors T n whose correlation with U is q, is qr. (2) We derive a formula for the probability that the sample correlation (over n subjects) using d is positive given that the true correlation is negative (and vice-versa); this probability can be substantial (around 20 - 30%) for values of n and q that have been used in recent affective computing studies. (3) With the goal to reduce the variance of correlations estimated by an automatic detector, we show that training multiple neural networks d(1) , . . . , d(m) using different training architectures and hyperparameters for the same detection task provides only limited “coverage” of T^n.