The ELCs illustrate how loud tones at every frequency need to be sound 'equally loud' - as loud as each other - to the ears of thousands of test subjects averaged together.
Using the graph curves, it suggests that we are most sensitive in the upper-midrange region, where reside the frequencies most critical to speech intelligibility. Conversely, less sensitive from 8kHz and higher frequencies, and MUCH less sensitive to lower midrange, mid-bass and deep-bass sounds(500Hz and down).
In fact, flip one of the curves in the middle upside down, and you'll have a good idea of the averaged frequency response of human ears. It's no wonder that more accurate(flatter) headphones and speakers sound so unappealing to many people, especially to average consumers who instead flock to headphones from Monster(Beats by Dre) etc.
Those are designer phones that cost up to twice as much as more accurate alternatives, and tend to have a more 'V'-shaped or smiley curve response. While initially appealing, that effect grows fatiguing after a while and does not serve the music being consumed through them.
Succinctly, it is better to seek out flatter headphones(Sony, Shure, and A-T all offer models for between U$100-130), and use the EQ built in to your device or the controls on your home listening system to accommodate your hearing preferences.