periments to repeat our findings might obtain morerobust results by using increased sample sizes. Afurther limitation of the field studies was that thenumber of respondents varied somewhat for eachmeasure. Some questions on the survey were acci-dentally skipped by the participants, leading to somemissing data for the self-report measures, while inother cases physiological data collection was not pos-sible because the chest-strap detector of the pulsemonitor did not always fit the participant sufficiently;some of the pace data were lost because the wrist-strap device failed to receive the data transmittedby the bicycle’s speedometer. We did not makeadjustmentsinthestatisticalanalysis,sincethemissing data were limited for most measures (seeTables IV and V). We should point out, however,that for the measure for which missing data weremost pronounced—the objective pNN10 measure—our ability to detect a significant difference wouldhave been reduced.Notwithstanding these limitations, those changesin experienced risk reported by participants were notreflected by changes in our selected measure of emo-tional challenge, pNN10. Even when we attemptedto increase the level of psychological challenge in afollow-up study, by asking participants to cycle with-out a helmet using one versus two hands, there wasno change in pNN10 despite increased differences incycle pace, discomfort, excitement, personal insecu-rity, and accident risk between conditions (Table VI).In both main and follow-up field experiments, how-ever, we could not be certain that any differences inpsychological challenge measured by the HRV in-dicator were not masked by variations in physicalload within or across participants, even though wehoped that the downhill gradient would minimize anysuch effects. Increased cycling speed in the absenceof a helmet was associated with increased heart rateand reduced pNN10, which implies that participantsvaried their speed using the pedals, rather than thebrakes as we had envisaged. We attempted to ac-count for the confounding effect of varying physicalload on pNN10 as a measure of mental load by multi-plying it by heart rate, but this new measure of emo-tional challenge was linked neither to self-reportedrisk nor helmet wearing.1Another explanation for the lack of observedchange in psychophysiological measures is that par-ticipant self-reports and cycling speed might havebeen the result of cognitive rather than emotionalprocessing.(32)In line with the “zero-risk theory” ofN¨a¨at¨anen and Summala,(12,33−35)helmet users mayhave kept full subjective control over their riding inboth conditions, such that no physiological responseconcomitant to a cognitively reasoned speed decre-ment occurred.