(1) When the teeth are together (required for clenching/grinding), the packet will rupture
This Assumption can be verified easily by the user when awake.
(2) When the packet ruptures, the substance released will be sufficiently disagreeable to wake or otherwise alert the user
There is effectively no limit to the disagreeable taste of liquid substances. Consider the use of capsaicin, the harmless "active ingredient" in tabasco sauce, jalapeno peppers, and “pepper spray”. Pure capsaicin is not a substance one would want to come into contact with, being approximately 5000 times hotter than Tabasco sauce. However, a suitably diluted solution that is safe will still be (temporarily) quite uncomfortable. A wide variety of substances are available that have equal potential. It is therefore likely that a suitable substance could be identified for any given individual. Additionally, if the user wakes with a broken packet and no recollection of the event, he or she will be aware that a stronger solution is needed.
(3) Upon breaking packets, the user has the presence of mind to replace fresh packets on the device, or use another device pre-loaded with packets
This may be the most difficult assumption to validate. The behavior of sleepy people is difficult to predict. However, there are strategies to help, such as producing a series of packets containing increasing concentrations of the chosen substance. A user could start with a mild solution and “dial up” the strength only as needed. Another strategy is to have another identical device available on the nightstand so that it is very easy to replace. Additionally, anecdotal evidence indicates that even if packets are not replaced, the frequency of bruxing (as experienced on subsequent nights) is still likely to decrease.
Taking these assumptions one at a time, they all seem very reasonable and defensible for a motivated bruxer. A fourth assumption makes the approach more appealing:
(4) The frequency with which the user breaks packets will decrease over time as the subconscious is reinforced that this behavior is immediately unpleasant.
Anecdotal data from the very few users that have been able to independently construct taste-based devices support assumption (4). The general pattern seems to be that for the initial period of usage a user may be awakened relatively frequently, followed by a decrease in bruxing, perhaps only breaking a packet every couple of weeks. Clearly a clinical test will be required to fully validate this assumption.