The human ear is tuned to be the most sensitive at bird-song frequencies. It also uses higher ordered harmonics to calculate how loud a sound is- its not doing that from the fundamental frequencies as such are quite rare in a pure form in nature. So if the system generates these harmonics (5th and above) even in trace amounts our ears are so sensitive to them that they will be heard, even though the distortion of the amp might be 0.005% THD.
The converts distortion into tonality. So the presence of these distortions is heard as both brightness and harshness (the 7th in particular having been known since the 1930s as a source of a metallic coloration), even though on the bench or in the room the system might measure perfectly flat.
In fact the ear has tipping points wherein tonalities created by distortion can be favored over actual frequency response errors. So in some cases its better to have none of these distortions rather than perfectly flat frequency response.
The converts distortion into tonality. So the presence of these distortions is heard as both brightness and harshness (the 7th in particular having been known since the 1930s as a source of a metallic coloration), even though on the bench or in the room the system might measure perfectly flat.
In fact the ear has tipping points wherein tonalities created by distortion can be favored over actual frequency response errors. So in some cases its better to have none of these distortions rather than perfectly flat frequency response.