"We" don't induce it. The MS software controls the source and can change the output to meet the required ion mode. In simple terms, the voltage is flipped. With sufficient stabilization, a multiplexer can be used to monitor both at "the same time". If you require a more detailed explaination, then please do a search for one of the many excellent tutorials on LC-MS operating modes.
BTW: For method development, and as long as it is safe to do so, it is a good idea to always monitor ions in both positive and negative ESI mode (Usually in a neutral mobile phase at first). Sometimes sample chemistry will surprise you and the other mode will reveal new peaks or show better response.
Indeed, Bill is right. The switch between ion modes is based on the change of voltage polarity on ESI capillary and ion optics of mass spectrometer. Nevertheless, you can significantly promote the formation of positive or negative ions adjusting solvent chemistry as you have already mentioned in your question. In the case of LC-MS such an approach also works good. For example, to obtain intense signals on positive mode you should preferably use acidic eluents and protonating solvents such as methanol instead of acetonitrile (it is most important for APCI mode). Of course, the variation of solvent and pH is limited in LC by stationary phase properties. In some situations, you can use alternative HPLC stationary phases withstanding the extreme pH levels such as Hypercarb. If it is impossible, the tuning of solvent chemistry can be done by post column in-line addition of appropriate mobile phase modifier (methanol or other solvents, acids, ammonia etc.).
@Cece Hong- that depends in part on the column used. Ammonium hydroxide is useful, but one needs to keep the pH below ~7.5 to 8 when using silica based columns. The silica dissolves in polar, basic, solutions. There are other "basic" mobile phases that can be used as well, so long as the buffer is volatile (such as ammonium acetate or ammonium formate).
Note the instructions that came with your column, some silica columns are modified to stand basic conditions.
Cece: Your questions sound a lot like homework assignments.... If you are just asking these questions out of curiosity or as part of a class assignment, then please read one of the many books on this topic for more information. LC/MS is a very complex analytical technique which has many possible ways of "doing things". It can not be summed up in a sentence or two.
*Yes, negative mode favors basic mobile phases (Jack provided some excellent advice above), but sometimes we even run NP mobile phases (i.e. Hexane/IPA !) in negative ESI mode and obtain enough signal for our samples w/o having basic conditions present.