===================================================================== ==> In this proposal, more time is being requested for AFRANCES.DUST_PG ===================================================================== How did galaxies form and evolve is still a mystery to a large extent. Conflicting observational results and theoretical predictions did not allow yet to decide in favour of either the long-standing paradigm according to which spheroidal galaxies have underwent a huge bust of Star-Formation (SF) at high redshift, making up the bulk of stars, or its current compeossiting scenario envisaging galaxy formation as a more gradual process of merging of smaller fragments taking place at more recent cosmic epochs. Recently detected high-z sources in the optical band invariably display low SF rates, more typical of forming disks than of young E/S0's. We have recently discussed two concepts (see Franceschini et al., 1994, ApJ427,140; Granato et al. 1996) that could help to solve this enigma. One is that our inability to detect forming high-redshift luminous spheroids might be due to the effect of extinction of the optical-UV emission by a dust-enriched ISM, in which case the optical emission would be deprived and most of the energy output would occurr in the IR. Furthermore, all forming spheroids probably harbor a (radio or radio-quiet) quasar, which would complicate further the issue (e.g. Franceschini and Gratton, 1996, MNRAS), but would also ease the target selection. Thanks to its ability to extend the optical exploration to substantially longer wavelengths (3-20 micron), ISO could substantially contribute to solve this enigmas. Two approaches will be followed to test these possibilities. One is to survey blindly entire portions of the sky to the faintest possible flux levels. The alternative one, adopted here, is to concentrate the efforts to study a limited number of candidates, including a few BAL QSO's, dusty radio-galaxies and dusty gravitational lenses. We propose to spend a few hours of the ISO mission to perform photometric imaging with the very sensitive CAM filters LW3 (or LW10) of 11 targets. This will also provide invaluable information on some mysterious objects in the distant universe.