The following principle is very plausible:
(*) One can only know what a pain is like if one has had a pain.
The principle is also plausible if one replaces "pain" with "pleasure", "auditory experience", "olfactory sensation", "pleasure", "a feeling of warmth", etc. However, (*) is false unless a certain version of disjunctivism is true. My argument for this shall apply to all the related principles as well.
Suppose we have Mary and Patricia. Mary is a neuroscientist who has never felt a pain. Patricia has had several physically painful, but non-traumatic, experiences in the past. She is not currently in pain. As long as Patricia remembers her painful experiences, we will say that Patricia knows what a pain is like. Moreover, let us suppose that it is not painful for Patricia to remember her painful experiences. It is typically painful to remember traumatic pains, and it is often painful to remember psychological pains, but it is often not painful to remember non-traumatic physical pains, and we can suppose that Patricia's are such. This supposition does not put into question Patricia's knowledge of what pain is like. We learn from these considerations that a version of (*) where "has had" is replaced with "is having" is very implausible.
Now, suppose that Mary implants in herself the memories of Patricia's painful experiences. If Patricia knows what pain is like on the basis of the memories, Mary should be able to know what pain is like on the basis of what she has just implanted in herself. Now, granted, what Mary has as a result of the memory implant are not strictly speaking memories, but apparent memories. If Patricia painfully fell of a bike at age 11, then Mary now has an apparent memory of painfully falling off a bike at age 11. But this memory is merely apparent because Mary did not in fact painfully fall of her bike at age 11. But it does not matter whether the memories be veridical or merely apparent for the knowledge of what pain is like. Mary knows that these are merely quasi-memories (to use Shoemaker's term), since she implanted them herself. But it seems plausible that they give her just as good an insight into what it would be like to have a pain as Patricia's genuine memories do.
Patricia knows by means of being able to see in memory what the experience of pain was like--without actually feeling the pain she is observing in memory. That the painful experience actually happened seems irrelevant to this, and if it is irrelevant to this, then it seems that Mary ought to be able to see in quasi-memory what the experience of pain was like. Subjectively speaking, after all, to remember and quasi-remember (or apparently remember) surely feels the same.
There is an objection to this argument, namely disjunctivism. If experiences feel differently depending on whether they are veridical or not (Ram Neta defends this view), then Patricia's memory and Mary's quasi-memory may not be similar enough for Mary to be able to know what the pain was like. But a disjunctivism about memory does not seem deeply plausible. Our memories are often foggy even when veridical. That they really do feel different when they are veridical does not appear all that plausible to me. But I have no solid argument against this disjunctivist thesis, and hence all I can say is that if this kind of disjunctivism is false, then one can know what pains are like without having had any.
Note: If the mental locally supervenes on the physical (at least for humans), Mary doesn't even need Patricia for this. She can just run a computer simulation of her brain, body and environment to figure out what her neural memory-correlate state (i.e., the neural state correlated with the memory of the event) would be several years after herself painfully falling off a bicycle, and then implant that neural state in herself.
Related question: Can God, without making use of the Incarnation, know what pains are like? The main reason to deny that apart from the Incarnation, God can know what pains are like is something like (*), combined with a claim that God doesn't feel pain qua God. My above argument, however, shows that (*) is probably false. And this undercuts the main reason for denying that God knows what pains are like. If one could show that my counterexample to (*) is the only kind there can be--so one must either be in pain or have a memory of pain or have a quasi-memory of pain to know what a pain is like--then one could rescue the argument against the possibility of God knowing what pains are like, since divine perfection is incompatible with non-veridical quasi-memories.