Thanks for your critique.
Hofstadter and others fail to take account of what happens when you have tens of millions of these loops, each with its own unique memory contribution at different abstraction levels for even a simple sentence like "that's the tree." That said, there will be simpler neural 'brains' that suffer such lack of memory simply because they cannot sustain rich enough memories through the sheer problem of lack of parallel computation quantity at every level of abstraction comparable to what large highly differentiated brains.
There is of course a large and good scholarly literature on consciousness.
I will let my claims stand. If you read the first referenced article and the articles referenced there you will get a clearer picture. Normally I have steered clear of consciousness claims precisely because it would be a full time job just defending the proposed theory. Look at what the theory predicts and how it predict it. Hofstadter is also not a language person (technically structural linguistics). My proposal places all memory in the diversity of predications that can be observed directly in precisely the predications that you have listed as counterexamples.
A further side effect of the present theory is that segments of neural tissue can be damaged (in any number of thousands of ways) and yet reconstruct memories as the hierarchicalization processes (whether linguistic or not) can reconstruct the memories. My Ph.D. is in verbal learning and memory on memory for causation (1976). A lot of claims make over simplistic assumptions based on unrealistic estimates of the number, diversity, multi modal and intermodal abstraction, of such 'loops' all in operation to what appears subjectively to be simple at any conscious moment.
I did like Hofstadter's book and do similarly recommend it as a reading in the history of computational cognitive neuroscience. I also highly recommend Sejnowski's "The Deep Learning Revolution" https://www.amazon.com/Deep-Learning-Revolution-MIT-Press/dp/026203803X
which poses the question of "where is memory" and which this theory answers by saying "it is dynamically constructed all the time in the dynamics and kinetics of neural systems observable directly in human natural language communication, brain to brain, in action."
Here is my review of Sejnowski's book.
https://medium.com/liecatcher/learning-deep-learning-bd883aa884b9
Here is a bit more critique based on a lack of credit given to what can be objectively observed now directed to the human addictive fascination with computer games....and how they consciously talk about games.
https://medium.com/liecatcher/deep-learning-devolution-in-the-evolution-of-gaming-7f3671aeb728
A final more modern excellent book if you haven't read it (although I suspect you have) is reviewed here on Mark Solms excellent work (regarding expressly your issues):
https://medium.com/liecatcher/off-on-mark-solms-3d87462ab327:
Note I make my explanation of consciousness as a prediction from a cognitive computation theory that never was designed to explain consciousness. It is a natural consequence. Here is the evolutionary claim that precedes the consciousness claim.
https://medium.com/liecatcher/the-natural-evolution-of-human-lies-655e983ee6c6
The present theory is further buttressed and predicted from the study of Mendaciolgy
https://medium.com/liecatcher/mendaciology-4e21bd47d8ba
Jeff Hawkins provide other good reads although like others he has no appreciation of observations that can be made based in the study of observable natural language behaviors.
