Note: This question arguably spans mathematics and linguistics, and possibly other disciplines, but I am posting it here because I think that mathematicians are best placed to answer it.
The base-10 (decimal) number system is the most common symbolic counting system used by people today. Scholarly literature on number systems state that this is derived from the use of fingers in the representation of numbers (see e.g., Ores 1948, p. 1-2). However, there are also cases of some cultures using other bases, up to base-20. It seems evident from this history that humans can handle number systems with various bases.
The use of binary numbers in electronic computing occurs because it is efficient to manufacture components that exist in a binary state. This suggests that the most efficient base is the smallest base, which is the binary system. Since this is (arguably) the most efficient system for electronic computation, it is arguably desirable for humans to be able to easily and rapidly convert between their own common number system, and the binary system.
This suggests that it would be more efficient for humans to adopt a number system that is base-2$^k$ for some $k \in \mathbb{N}$. The obvious choices, operating within an appropriate range of symbols and digits, are the octal system (base-8) or the hexadecimal system (base-16). These systems can be trivially converted back and forth with binary numbers, since they merely require the user to know the binary representation of each base element, and string numbers together with the place method.
Question: Aside from its derivation from finger counting, is there any property of the base-10 number system that gives it a natural advantage over other bases for human use? What are the pros and cons of adopting a base-2$^k$ number system (e.g., base-8 or base-16) in place of the presently common decimal system? Would conversion to one of these systems be desirable?
Some preliminary thoughts on the pros and cons of the decimal number system (hopefully this can be supplemented by other users):
Finger counting: As noted in the question, decimal corresponds to finger counting, so it is very simple for people to count decimal numbers on their fingers. It would be more difficult (but not impossible) to count on other bases on the fingers. Bases higher than ten become more difficult, which militates against a base-16 system.
Division by simple primes: It is desirable for the base of the number system to be a product of simple prime factors, so that the base is divisible by simple primes. The decimal number system uses base $2 \cdot 5 = 10$, which uses the first and third prime. This makes it easy to divide by these two simple primes. This is arguably less simple than a base-2$^k$ number system, which can easily be divided by the smallest prime multiple times. Nevertheless, the decimal system allows easy division by five, whereas the base-2$^k$ systems do not.
Inertia/conversion costs: Given that the decimal system is in such widespread usage there would be substantial costs to converting numbers to a base-2$^k$ system. There would be costs in time and frustration for people to learn and get used to the new system, and there would be monetary costs to the translation of documents, etc., which represent numbers in the decimal system. These costs would be very large, which militates in favour of using the existing decimal system. On the other hand, the conversion costs are likely to grow exponentially over time (as new documents are generated in the decimal system) and so if it is desirable to convert, then it is probably preferable to convert earlier rather than later.