Dust grains are an important component of star forming regions and are therefore a powerful tracer of the location and mass of prestellar cores. However, the properties of dust grains in such dense regions are poorly constrained, causing difficulties in the modeling of the physical properties of dense cores. One of the fundamental properties of dust grains is their size distribution, which is relatively well-known in the diffuse ISM but not very well constrained in dense regions. Indirect evidence is consistent with the presence of large grains; however, to date, interpreting submm emission measurements or absorption in the near-infrared (NIR)/mid-infrared (MIR) domains is difficult due to the ambiguity between density (and temperature if in emission) variations and grain properties. Recently, we have discovered emission at 3.6 and 4.5 μm towards the densest parts of L183. We have named this effect coreshine, by analogy with the cloudshine, seen at the surface of dark clouds in the near-infrared. This 3.6 and 4.5 μm emission can only be explained by the strong scattering of background interstellar radiation due to micron-size grains and show that the average grain size increase needs to be proportional to the cloud core density gradient to reproduce the observations. We show that the scattering effect is very sensitive to the grain size distribution; therefore these observations will provide a new tool with which to study grain properties. In a subsequent search through about a hundred of dense low mass core regions in the Spitzer Heritage Archive, we find that the effect is present in  ~50% of the sources, including prestellar cores, Class 0 and I sources with or without outflow, showing that the coreshine effect should rapidly become a general observational method to investigate dark cloud and grain properties.