What are the appropriate limits when criticizing the previously published work of others that you're citing in a paper? Are there limits with respect to maintaining decorum and decency?
Generally speaking, any critiques should be brief and directed (i.e., "the cited works failed to consider <X> in their analysis", "to simplify their model, they assumed <X>, whereas in reality <Y>").
If you're criticizing assumptions, be sure to specifically state which assumptions you challenge, and clearly deliniate (with references) why they are incorrect. Note that many papers use incorrect assumptions to begin work in a new field, and your critique should recognize that (i.e., "The seminal paper assumed <X> . We extend this work by removing said assumption.")
If you're criticizing conclusions, again state specifically which conclusion you disagree with, and be sure to provide data/analyses to back up your conclusion.
If you're challenging their data, note that (in neuroscience, at least) this is perfectly common; findings differ all the time. Progress often stems from finding the cause of these differences between data. You should note the difference and mention something along the lines of "Our findings differ from those of <X> , and we believe this is because <Y>". Note that it is almost a requirement to mention something like this; if you don't, (1) the researchers whose paper you missed will probably mildly insulted, and (2) people familiar with the field will assume you didn't do your basic literature search, which makes you look stupid.
Needless to say, ad hominem attacks are always inappropriate and should never appear in scientific literature.
I think that simply by thinking in terms of "criticism" you have already gone too far. I try and point out what the previous work has done and demonstrate how my work builds on that. Things like reducing the number of assumptions, doing additional analysis, or using a better method, don't require you to criticize the previous work.