There are two different transporters for T3 and T4 into the brain. One (OATP1c1) transports only T4, the other (MCT8) transports T3 and T4. T4 is then converted locally to T3 by the D2 deiodinase enzyme. The total T3 in the brain comes from what was converted locally (from T4), plus what was transported in as T3. 
Deiodinase activity is different in specific regions of the brain. Thyroid hormone levels in the brain are kept in tight ranges because the brain requires that stability. In hyperthyroidism (when T3 levels are too high), D3 expression increases, which increases the T4 to reverse T3 pathway, and D2 expression is suppressed, lowering T4 to T3 conversion. These two processes work together to lower high T3 levels. Likewise, in hypothyroidism (when T3 levels are too low), D2 expression is increased, which raises T4 to T3 conversion, to raise T3 levels in the brain. When someone takes T3-only and no T4, they may lose this important regulatory feature of reverse T3 (to lower high T3 levels), and T3 levels may exceed the brain’s optimal range, since there is no T4 to inactivate. It’s analogous to running too much voltage through a low-voltage appliance. This can result in hyperthyroid dementia or fresh amnesia, where recall of events minutes earlier is impaired. [40,59] I suffered from this fresh amnesia and could not recall things I’d done two minutes earlier. I was also completely unable to perform simple math in my head, which I do routinely. My memory and math skills returned once my T3 dose was reduced. In another case, the person lost their foreign language fluency when they took too much T3.
A study of rats treated with T3 supports these observations. After T3 treatment, their brains showed a 50% decrease in specific membranes of the cerebral cortex, which is the part of the brain involved with memory, attention, and language. 
Thyroid hormone levels of healthy women (all thyroid labs within normal range) were analyzed for any correlation of their performance in neuropsychological tests to their thyroid levels. High Free T3 levels were positively associated with slower completion times for certain cognitive tests; in other words, the higher the Free T3, the longer it took the subject to complete three tests. In Trail Making Test-A, subjects draw a line connecting numbered circles, which are randomly placed on the page. In other words, after finding circle 1, they draw a line to circle 2, etc. until they reach circle 25. In Trail Making Test-B, subjects have to connect both numbers and letters in order: 1-A-2-B-3-C etc. This test can cause extreme confusion and will take longer to complete if the brain isn’t working correctly. The third test that subjects with high Free T3 took longer to complete is the Tower of London test. In this test, subjects rearrange stacked, colored beads on three posts into a new configuration they’re shown. A bead that was formerly on the bottom of a post in the old configuration may need to be on the top of a different post in the new configuration. Getting the beads in the correct order involves thought and planning. Because high Free T3 was consistently correlated with slower performance test times, the authors concluded that elevations in thyroid hormones (within the normal range) may negatively affect frontal cortex executive functions in the brain, where memory, attention and language are located. 
Female thyroid cancer patients who had undergone thyroidectomies, had radioactive iodine, and were on suppressive doses of levothyroxine were studied when their levothyroxine doses were to be stopped for an upcoming radioactive iodine whole body scan. They were tested at three points in time:
- before stopping their dose, when they were considered mildly hyperthyroid (suppressed TSH, over range FT4, normal to high FT3)
- 4-7 days after stopping their dose, when considered euthyroid (FT3 and FT4 normal, but TSH still suppressed for most), and
- about 30 days later, when considered profoundly hypothyroid (TSH high, FT3 and FT4 below range). Controls of similar ages and backgrounds were also measured for comparison.
A visual scanning test that measures distractibility and visual inattentiveness gave interesting results. Subjects had to locate a particular symbol in a paper filled with a matrix of symbols, of which 60 were correct. While nearly all symbols were found by all groups (59/60), the subjects took the longest time to perform the task when they were mildly hyperthyroid, longer even than when they were profoundly hypothyroid!  This suggests that too much thyroid hormone causes some type of brain dysfunction, and may actually be a mild form of hyperthyroid dementia. In fact, Graves’ hyperthyroid patients frequently exhibit deficits in attention, memory, and complex problem solving. 
The hippocampus and temporal cortex areas of the brain, which affect memory and cognitive functions, exhibit the highest D3 concentrations, the enzyme that inactivates T3. This suggests that these two areas of the brain are the most sensitive to high T3 levels, and the studies just mentioned seem to confirm that statement. In one experiment, D3 could not be detected in hypothyroid brains, and D3 levels were found to correlate with thyroid status in the central nervous system. In other words, D3 would rise as thyroid levels rose, so reverse T3 could be made if T3 levels became too high. [29,30]
Brain size physically changes when thyroid levels are too high or too low. The hyperthyroid brain is larger than normal, and the hypothyroid brain is smaller than normal. Conversely, ventricular size (which contains the cerebrospinal fluid) is smaller than normal in hyperthyroid patients, and larger than normal in hypothyroid patients. The reduction in brain size and increase in ventricular size in hypothyroid patients significantly correlated to reduced T4 levels. In other words, as T4 levels rose, brain size increased and ventricular size decreased towards normal. But the correlation of brain size with T3 levels was not statistically significant. However, ventricular size decrease strongly correlated with rising T3 levels.  It is possible that the rapid decrease in ventricular size in the brain might have something to do with the headaches some patients report when increasing their T3 dose on the T3-only protocol.
Another study found that locally deiodinated T3 (from T4 in the brain) accounted for more than 80% of the total T3 specifically bound to nuclear receptors in the cerebral cortex, and approximately 67% of that in the cerebellum. T4 would then be the major source of intracellular T3 in the central nervous system.  If 80% of the T3 in the brain came from T4, then a higher than normal dose of T3 would be necessary to compensate for this loss of T4 in a T3-only protocol. But a dose that high may have adverse effects on other systems that are more sensitive to T3, like the cardiovascular system, and may cause problems like tachycardia (rapid heart rate) and high blood pressure.
EEGs (brain electrical activity) of hypothyroid patients (thyroidectomy followed by radioactive iodine) were compared while they were still hypothyroid, after supplementation with T3, and after supplementation with T3 + T4. The EEGs only normalized when T4 was added to the T3, and correlated significantly with the rise in serum T4 levels. It appears then, that T4 is essential for normal brain function. [32,33]
The psychological well-being of patients on thyroid hormone replacement was compared to their Free T3, Free T4, and reverse T3 levels. There was a strong positive correlation of higher FT4 levels with well-being; in other words, patients with higher FT4 levels (even above the reference range) just felt better mood-wise. But there was no correlation of psychological well-being to Free T3, rT3, rT3/FT4, or FT3/rT3. The authors noted that all measures are serum measures and do not reflect intracellular levels, and that many tissues obtain their T3 by conversion from T4.  As stated earlier, with ample T4, the brain can create the optimal amount of T3 it needs with the appropriate deiodinase enzymes. If T4 levels are too high, more will be shunted to rT3. If T4 levels are low, nearly all T4 will be converted to T3. But if T4 is extremely low, then T3 in the brain will be insufficient.
28. Robertas Bunevicius and Arthur J. Prange Jr. “Thyroid-Brain Interactions in Neuropsychiatric Disorders” in Neuropsychiatric Disorders, edited by K. Miyoshi et al., 2010.
29. Tu HM, Legradi G, Bartha T, Salvatore D, Lechan RM, Larsen PR. Regional expression of the type 3 iodothyronine deiodinase messenger ribonucleic acid in the rat central nervous system and its regulation by thyroid hormone. Endocrinology. 1999 Feb;140(2):784-90. http://www.ncbi.nlm.nih.gov/pubmed/9927306
30. Santini F, Pinchera A, Ceccarini G, Castagna M, Rosellini V, Mammoli C, Montanelli L, Zucchi V, Chopra IJ, Chiovato L. Evidence for a role of the type III-iodothyronine deiodinase in the regulation of 3,5,3′-triiodothyronine content in the human central nervous system. Eur J Endocrinol. 2001 Jun;144(6):577-83. http://www.ncbi.nlm.nih.gov/pubmed/11375791
31. R. Crantz, J. E. Silva, P. R. Larsen. An Analysis of the Sources and Quantity of 3,5,3′-Triiodothyronine Specifically Bound to Nuclear Receptors in Rat Cerebral Cortex and Cerebellum. Endocrinology. February 1, 1982 vol. 110 no. 2367-37. http://endo.endojournals.org/content/110/2/367.abstract
32. Pohunková D, Sulc J, Vána S. Influence of thyroid hormone supply on EEG frequency spectrum. Endocrinol Exp. 1989 Dec;23(4):251-8. http://www.ncbi.nlm.nih.gov/pubmed/2620656
33. Bauer, M., Goetz, T., Glenn, T. and Whybrow, P. C. (2008), The Thyroid-Brain Interaction in Thyroid Disorders and Mood Disorders. Journal of Neuroendocrinology, 20: 1101–1114. http://onlinelibrary.wiley.com/doi/10.1111/j.1365-2826.2008.01774.x/full
40. Toshiya Fukui, Yukihiro Hasegawa, Hiroki Takenaka. Hyperthyroid dementia: clinicoradiological findings and response to treatment. Journal of the Neurological Sciences – 15 February 2001 (Vol. 184, Issue 1, Pages 81-88). http://www.jns-journal.com/article/S0022-510X%2800%2900487-1/abstract
59. Ii Y, Ohira T, Narita Y, Kuzuhara S. [Transient dementia during hyperthyroidism of painless thyroiditis. A case report]. [Article in Japanese] Rinsho Shinkeigaku. 2003 Jun;43(6):341-4. http://www.ncbi.nlm.nih.gov/pubmed/14503353
60. Oatridge A, Barnard ML, Puri BK, Taylor-Robinson SD, Hajnal JV, Saeed N, Bydder GM. Changes in brain size with treatment in patients with hyper- or hypothyroidism. AJNR Am J Neuroradiol. 2002 Oct;23(9):1539-44. http://www.ncbi.nlm.nih.gov/pubmed/12372744
62. Orford MR, Leung FC, Milligan G, Saggerson ED. Treatment with triiodothyronine decreases the abundance of the alpha-subunits of Gi1 and Gi2 in the cerebral cortex. J Neurol Sci. 1992 Oct;112(1-2):34-7. http://www.ncbi.nlm.nih.gov/pubmed/1469437
65. Grigorova M, Sherwin BB. Thyroid hormones and cognitive functioning in healthy, euthyroid women: a correlational study. Horm Behav. 2012 Apr;61(4):617-22.
66. J I Botella-Carretero, J M Gala´n, C Caballero, J Sancho and H F Escobar-Morreale. Quality of life and psychometric functionality in patients with differentiated thyroid carcinoma. Endocrine-Related Cancer (2003) 10 601–610. http://erc.endocrinology-journals.org/content/10/4/601.full.pdf
67. Trzepacz PT, McCue M, Klein I, Levey GS, Greenhouse J. A psychiatric and neuropsychological study of patients with untreated Graves’ disease. Gen Hosp Psychiatry. 1988 Jan;10(1):49-55. http://www.ncbi.nlm.nih.gov/pubmed/3345907
68. Saravanan, Ponnusamy, Theo J. Visser, and Colin M. Dayan. Psychological well-being correlates with free thyroxine but not free 3, 5, 3′-triiodothyronine levels in patients on thyroid hormone replacement.Journal of Clinical Endocrinology & Metabolism. 91.9 (2006): 3389-3393. http://jcem.endojournals.org/content/91/9/3389.long