Nick is correct when he says that the EQ is used to cut frequencies which mask other instruments, though the general idea of multi band EQ is to leave the signal at its original total level- so that the overall effect is neither cut nor boost, when taken across all frequencies. Cut 6dB here, boost 6dB there. (that is a lot, by the way- more subtle is usually better)
One mistake which is often made is to listen to individual instruments and tweak the EQ until each one, on its own, sounds how you think it ought to. Mixing them all together, the result is always overlapping frequencies, and a mushy lack of separation. The guideline is that you don't know what the individual instruments should sound like until you have heard the ensemble.
Actually there are several ways of achieving separation. Instruments can be heard as separate if they are separated by:
-Space: position in the stereo field, or sometimes in the reverb soundfield- if it sounds further away, its not the same instrument!
-Tone: by natural or processed EQ
-Pitch: you never confuse a double bass with a piccolo
-Attack: an instrument with a percussive attack like a guitar, will sound separate from one with a breathy attack like a flute, even when playing in unison.
-Time and rhythm: two instruments playing one after the other, call and response style, or one instrument playing quarter notes (crochets) will sound distinct from another playing sixteenths (semiquavers).
Hope this helps,
Cheers
Dave