Evolutionary reliabilism (ER) is the view that natural selection likely favored reliable cognitive faculties in humans. In his 2004 paper, James Sage (PhD, University of Utah) argues that there is no reason to accept that natural selection favored truth-reliable.
Donald D. Hoffman (University of California, Irvine) argues that true perception should not be assumed to have been favored by natural selection.
He was interviewed in Closer to the Truth to speak about his work. In this video, he gives piratical examples of evolution preferring reproductive fitness over truth.
Here is a complete list of Huffman’s research.