Cross-individual integration  of  information  has  more  than half a Century of research history in the context of joint decision-making. In group psychophysics formal models have been developed to describe  processing  of information  during  group  interactions in perceptual tasks and explain why groups systematically fail to integrate available information optimally. We propose  a distributed averaging model which allows for efficient (and optimal  under certain conditions) integration of individual judgments with minimal  data  exchange.  To test the model, we developed a new paradigm,  where participants were asked to match locations on the two planes  of  a  virtual environment. In a series of experiments we show how  participants  could  flexibly  incorporate information from their collaborators,  and that cross-individual integration was as efficient as within-individual   one  even  without  exchange  of  uncertainty information between individuals.