Metal detectors can, in their basic form, discriminate between ferrous and non-ferrous metals. Basically, a ferrous metal will normally tend to increase the inductance of the "search-coil" whilst non-ferrous metals will reduce the inductance of same coil.
The most basic metal detector is a BFO type and its search-coil inductance forms a tuned circuit oscillator with a parallel capacitor - the oscillator frequency rises in the presence of non-ferrous metal and usually falls with ferrous metal.
Inductive balance metal detectors tend to be the main type used by beachcombers and these have a fixed oscillator coil and two "outer" receive coils that are wired antiphase producing zero net signal in the presence of no metal. The outer winding signals get amplified and used as the means of detecting metal. Like a BFO, inductive balance detectors can discriminate between ferrous and non-ferrous metal by the phase shift of the signal received.
Simple metal detectors (to discriminate against nails and other non-significant ferrous metal) will force an imbalance in the coils electronically and, ferrous metal will decrease the imbalance-signal whilst non-ferrous (likely to be more valuable) will increase the imbalance-signal. This will trigger a buzzer or light to indicate to the "user" the presence of a non-ferrous metal.
That's about as far as it goes on beachcombing detectors but the food and pharmaceutical industry have gone a step further. The received signal is processed and two parameters are extracted. These are the resistive and reactive components of the received signal and can tell you a fair bit about the make-up of the metal.
In these more sophisticated metal detectors, it is easy to discriminate between (say) 1mm iron and 1mm aluminium or 1mm brass - each will have there own signature relationship (resistive to reactive relationship) and, the overall amplitude can tell you how close the "foreign" object is to the search coils. It's fairly easy to recognize the difference between 1mm Fe and 2mm Fe based on phase angle alone - the amplitudes could be similar if the 1mm metal was close to the search head compared to the 2mm metal but there will be something like a 10º phase angle difference at a running frequency of 300 kHz. Different frequencies can be used to exploit different things.
Unfortunately there is some overlap in phase angle (resistive to reactive relationship) between different metals at different sizes but this can usually be overcome by knowledge of the overall amplitude.
Will this technique be able to tell the difference between aluminium and iron? Yes
Will this technique discriminate between iron and titanium? Almost certainly
Will this technique be able to recognize a regular and repeatable mass of aluminium and discriminate it from titanium? Probably, but tests would need to be carried out to see how easy that is, given the likely signals from various metal types.