Another important step forward in the exploration of the galaxy population at high redshift came with the development of large format near-infrared (NIR) detectors. Deep, wide-field surveys in the K band led to the discovery of a class of faint galaxies with extremely red optical-to-NIR colors (
). Follow-up spectroscopy has shown that these extremely red objects (EROs) typically have redshifts in the range 
. There are two possible explanations for their red colors: either they are galaxies dominated by old stellar populations with a pronounced 4000 Å break that has been shifted red-wards of the R-band filter, or they are starbursts (or AGN) strongly reddened due to dust extinction. Spectroscopy of a sample of 
suggests that they are a roughly equal mix of both (Cimatti et al., 2002a).
Deep imaging in the NIR can also be used to search for the equivalent of 'normal' galaxies at 
. As described above, the selections of LBGs, LAEs and sub-mm sources are strongly biased towards systems with relatively high star-formation rates. Consequently, the population of high-redshift galaxies picked out by these selections is very different from the typical, present-day galaxies whose light is dominated by evolved stars. In order to select high-redshift galaxies in a way similar to how 'normal' galaxies are selected at low redshift, one has to go to the rest-frame optical, which corresponds to the NIR at 
. Using the InfraRed ExtraGalactic Survey (FIRES; Labbé et al., 2003), Franx et al. (2003) identified a population of galaxies on the basis of their red NIR color
, where the
and 
filters are similar to the classical J and K filters, but centered on somewhat shorter wavelengths. The galaxies so selected are now referred to as distant red galaxies (DRGs). The color criterion efficiently isolates galaxies with prominent Balmer or 
, and can therefore be used to select galaxies with the oldest stellar populations at these redshifts. However, the NIR color criterion alone also selects galaxies with significant current star formation, even dusty starbursts. The brightest DRGs (Ks < 20) are among the most massive galaxies at , with stellar masses 
likely representing the progenitors of present-day massive ellipticals. As EROs, DRGs are largely missed in UV-selected (e.g. LBG) samples. Yet, as shown by van Dokkum et al. (2006), among the most massive population of galaxies in the redshift range 
, DRGs dominate over LBGs both in number density and in stellar mass density.
Using photometry in the 
and K bands, Daddi et al. (2004) introduced a selection criterion which allows one to recover the bulk of the galaxy population in the redshift range 
, including both active star-forming galaxies as well as passively evolving galaxies, and to distinguish between the two classes. In particular, the color criterion 
is very efficient in selecting star-forming galaxies with 
, independently of their dust reddening, while the criteria 
predominantly passively evolving galaxies in the same redshift interval. At 
-selected star-fo galaxies typically have higher reddening and higher star-formation rates than UV-selected ies. A comparison of 
galaxies with DRGs in the same redshift range shows that many DRGs are reddened starbursts rather than passively evolving galaxies.
