[tacgunner1]

During a psychology class, I came across a graph displaying the sensitivity to frequencies humans posses. 

It was similar to the attached graph. As our sensitivity to high and low frequencies is lesser, we would thus (to the extent of my knowledge) require a superior amplitude of these frequencies to compensate. However, I observed that on practically all headphones frequency response charts I have seen, the lower frequencies are either linear or below the same intensity as the mid-range, and high frequencies drop off as well. Would that not mean that the lower and higher frequencies would be almost a whisper in contrast to the mid-range due to our lesser perception of these?  I do not know wherein the issue lies: with my perception of how things SHOULD work, or if there is a factor I ignored. Could someone please enlighten me? 

 

[wakibaki]

The transducer response is linear across the range as far as the manufacturer can make it. That goes for both microphone and headphone or speaker.
 
So a piece of music is recorded with equal weight given to all frequencies. When it is played back an equal weight is given to all frequencies.
 
Hence the playback sounds (and is effectively) identical (!) to the original to the listener.
 
The fact that the listener doesn't respond equally to all frequencies (can't hear some) is irrelevant.
 
w

 

[mikeaj]

Exactly, real life sounds are affected by the unevenness of the aural response, so you want sounds played back to also be subjected to them (which they are; you don't want to compensate for it because uncompensated, it matches real life). That is, assuming playback levels are matching the live sound levels.
 
If they aren't and you're listening significantly quieter where the shape of the curve of the graph is different, the response might seem a little bit off, in addition to the very glaring and primary reality that it's significantly quieter. In these cases, usually it's harder to hear the bass, which shouldn't be surprising if you take a quick look at the graph.

 

[bigshot]

the only time when you need to apply compensation curves is when human ears aren't involved, either in mixing and mastering the music or in receiving the sound from a transducer.

Humans automatically compensate. No graphs or diagrams necessary. Just ears.

I prefer to adjust frequency response curves by ear for two reasons... 1) ears are what I listen to music with, and 2) if I can hear it, it matters to me and if I can't it doesn't.