This paper reports the results of an experiment that sought to establish if the parameters in an existing F0 implementation model uniquely correspond to perceptual dimensions like 'emphasis' and 'liveliness'. These results are negative. A discussion of the rationale behind different kinds of perceptual experiments into perceived prominence leads to the conclusion that such experiments may reveal the effects of contour-internal relations like downstep and declination, or may more generally reveal the existence of various aspects of canonical contour shapes, but will not yield a calculus for determining the perceived prominence of pitch accents on the basis of anchor points in the surrounding unaccented parts of the contour.