Cognitive EQ and the Fidelity Problem: Representational versus Experiential Fidelity in Algorithmic Curation

Algorithmic content curation faces a design question that is rarely posed explicitly: what should a feed be faithful to? Current platforms optimise engagement and treat the downstream psychological effects as unintended side effects. This paper argues that the implicit choice is between two distinct fidelity concepts. Representational fidelity surfaces content as it exists in the system, without compensating for known biases in posting behaviour or in how content is processed. Experiential fidelity compensates for those biases to produce accurate psychological calibration, even where this requires showing non-representative content. The distinction is illustrated with the Nordschleife principle from racing simulation: a flat screen cannot convey a 15 per cent gradient as embodied experience, so a designer must choose between modelling the gradient accurately and exaggerating it to 25 per cent so that the mediated experience matches the unmediated one. Neither choice is neutral. Applied to social media, if negativity bias makes negative content roughly four times as impactful, a representationally accurate feed produces an experientially distorted calibration. We propose user-selectable calibration modes, named Grateful, Dopamine, Accurate, Compensated, and Humble, as a design solution that sidesteps paternalism by giving users authority over their own calibration targets. In an era of synthetic media that can scale infinitely and where disclosure may not reach the level at which calibration occurs, architecture is a more effective intervention point than content labelling.