Strawson's "dot" model of co-referential thinking is anti-Fregean because:
(1) Thinking of an object as the same involves eliminating multiple modes of presentation
(2) A single dot can equivocate in its reference (so sense doesn't determine reference)
Millikan endorses the dot model and argues that Evans's dynamic thoughts are equivalent to the dot model.
Worries about the argument: there are functionally significant reasons for hanging on to multiple modes of presentation for an object, rather than eliminating them when you learn they co-refer. You might, e.g., want to hang on to a variety of modes of presentation for an object because other people refer to the object by way of those modes and don't realize they co-refer.