Eating: Close enough. A fetus drinks. And, in context, "eat" is an often-used colloquialism I didn't think would be misunderstood, since fetuses consume food (had I thought to write that instead of "eat", I would have thought it too cumbersome for here, even from me), but, apart from use of the umbilical cord, "[b]aby is able to suck and swallow" in week 17 of a pregnancy and in week 18 "[t]he baby is swallowing amniotic fluid" (https://www.hopkinsallchildrens.org/community/fit4allmoms/your-pregnancy-week-by-week) and "AF ["[a]mniotic fluid"] contains carbohydrates, proteins and peptides, lipids, lactate, pyruvate, electrolytes, enzymes, and hormones", thus having nutritional value (http://www.nature.com/jp/journal/v25/n5/full/7211290a.html). The quantity being swallowed in the third trimester is "up to a liter a day" (https://www.psychologytoday.com/articles/199809/fetal-psychology), up to about a quart a day.
What "unethical experiment" (@beaglelady)? General advice is to breastfeed but there are exceptions when formula is healthier, like if the mother is ill, and breast milk does not have enough vitamin D. So, depending on the health needs of neonates, a formula might be medically appropriate, and an experiment in which neonates choose between formulae that are all healthy would hardly be unethical, even accepting that neonates are more sensitive to some stimuli than are adults. And if an experiment would be unethical today, I think ethics were formally, articulately, and consistently introduced by outside authorities or committees into the approval process for experiment designs only in recent decades, and the findings of many older experiments can still be relied upon. Natural experiments (such as a survey of cliff-climbers who fell and probably should have died, given the circumstances) are not subject to the same ethics and their findings can still be accepted. As far as I know, no experiment I referred to in this topic would have been unethical even if it were done today, and I explicitly include the study of neonates eating from differently-colored containers. If you see that study as unethical, or any other in this thread, please tell us how.
Whether a baby with Tay-Sachs disease "[n]ever becom[es] . . . aware of its surroundings" (@aleo) was a point I didn't find agreement with among more-reliable sources. See https://ghr.nlm.nih.gov/condition/tay-sachs-disease, https://www.genome.gov/10001220/learning-about-taysachs-disease/ (e.g., "cognitively impaired" is not unawareness) (die "by age five"), https://www.mayoclinic.org/diseases-conditions/tay-sachs-disease/basics/definition/con-20036799, & https://medlineplus.gov/taysachsdisease.html (die "by age 4") (all as accessed 11-8-17). The view does seem to turn up in popular or nonprofessional discussions, and it may be the same view repeated in multiple places. It's possible that a baby might be born unaware of its surroundings and that would be relevant to this thread, but apparently that unawareness woould not be due to Tay-Sachs.
A fetus might "begin to learn language": "There is . . . a debate as to whether or not reading to your fetus can improve his or her brain development by actually helping him or her begin to learn language. While there is evidence that this is true, there is also evidence that all that the infant learns in the womb is lost after 21 days, so how much reading improves brain development is questionable." (Https://my.vanderbilt.edu/developmentalpsychologyblog/2013/12/the-do%E2%80%99s-and-donts-of-pregnancy-a-pregnant-womans-guide-to-birth-a-healthy-baby/.) While I disagree that science says that everything learned is lost so soon, although some probably is, this, in support of beginning to learn language, counters a view stated hereinabove. It appears that a fetus can distinguish words, in the way that we adults mean words in language, even though they may not know what they mean; they like a native language over a foreign language (https://www.psychologytoday.com/articles/199809/fetal-psychology), probably meaning familiar words over unfamiliar ones (familiarity could come from repetitive use), although the words when separated are probably still not significant to a fetus.
Whether symbols need to be "arbitrary" (@Christy) is not my opinion, even though they'd be abstract. Historically, symbols often start out have something in common with what they mean. It's likelier there'll be more in common between signifiers and signifieds when a biocommunication system and symbols are first being created. Insofar as that's true, arbitrariness would be reduced and possibly nonexistent.
I started to research for sources on whether someone can know of a signified without having a signifier for it. An initial glance through Google results suggests not, but search engine results page snippets are not good enough for a question like this. But I also still have not figured out an example of knowing of something (a potential signified) without having a signifier for it. If someone knows enough of a potential signified to struggle to create a signifier for it, the person actually already has a signifier for it and they're just trying to improve their signifier.
Knowing of something seems to have been questioned on the basis that thought is needed for knowledge and a fetus can't have thought yet. If thought is a higher-order property or quality than is knowledge, I would argue only for knowledge of a potential signified, not necessarily a thought of it. A prerequisite for a symbol therefore would be a capacity for knowledge, but that would seem to exist very early in development, perhaps prefetally and if not prefetally at least early enough to allow for the building of a collection of signifiers.
The fetus can see starting at 27 weeks (https://www.hopkinsallchildrens.org/community/fit4allmoms/your-pregnancy-week-by-week). I didn't know that. I didn't know there was any light to see inside, but apparently there's at least a little light and shadow. So it has even more potential signifieds than I thought.
It also hears, including the mother's speech and that of other people (https://www.psychologytoday.com/articles/199809/fetal-psychology). That, too, means more signifieds are potentially present.
"As the [7th] week progresses, the brain will divide into three parts: forebrain, midbrain, and hindbrain. The forebrain will be responsible for things such as reasoning, problem solving, and forming and retaining memories. . . ." (.) I don't know if the author meant that "reasoning, problem solving, and forming and retaining memories" occur in the fetus or only some time after birth.
I, too, have a memory that I have always ascribed to being from before birth, although I think it is of the seconds before birth or during the exiting process when light reflected off my surroundings that I came out with and ending when I quickly shut my eyes to the now-bright light. The memory includes an enhancement (little faces) that I may have applied at the time rather than later. While the comment above about a pre-birth memory seemed facetious, I gather some other people have reported pre-birth memories.
Fetal temperament and personality seem to have been accepted (https://www.psychologytoday.com/articles/199809/fetal-psychology).
The list of intrafetal symbols that I've thought of should, it seems, now be longer, given some of the known details of fetal development (see, e.g., https://www.hopkinsallchildrens.org/community/fit4allmoms/your-pregnancy-week-by-week & https://www.psychologytoday.com/articles/199809/fetal-psychology).
An argument that fetuses respond to stimuli without choosing becomes harder to sustain the more complicated the set of stimuli are and the less certain we are of how a fetus will respond to stimuli even after many observations. Eventually, that argument blends into someone never having choice, that everything even adults do is automatic. We should be careful about assuming that choice exists only when a possible chooser can tell us enough to convince us that they're choosing (a problem with addiction theory as applied to adults). No one said that in this thread but it is a common theme and it seems to be reflected in some statements in this thread and it often has its grounding in theology. It took a long time for scientists publicly to say that humans are animals in a good way and that nonhuman animals have qualities previously recognized only in humans. I understand that the problem may occur here, too, and scientific standards should be rigorous, but, even within that rigor, I think we have enough to go on in order to either investigate or conclude regarding various questions with respect to fetal capabilities being beyond what we used to believe. Science can be wrong (Einstein was wrong at least once or twice on important matters) but the way to correct scientific conclusions, which are always tentative, is not to go back to older beliefs on the ground that science might be wrong but to go forward by applying scientific method to refine or overturn intermediate scientific findings and thereby discover or confirm that science was wrong.
There's a rule, partly said tongue-in-cheek, called the Harvard rule: Given the right conditions in a properly-configured laboratory, rats will do as they ■■■■ well please. Given what we know of their culture, rats likely sometimes make choices. Can we be sure fetuses could never make choices? Or are some of us just saying that we don't know if fetuses make choices, not that scientists know that rats and fetuses never do? Any of us saying we don't know would be reasonable, even if we disagree. But lack of proof that something is true does not make it false; it may merely be unknown and fit for exploration.
(All URLs were as accessed 11-7-17 except as otherwise stated.)