Some thoughts:
*Descent with variation
I have long felt that "mutation" is not really the best term to use when explaining the idea of evolution. It conjures a mental image of conditions that are rare, alarming, and malfunctional. The truth is that every new individual carries variations in their gene pattern. The process of creating a new organism is long and complex when looked at from the level of acids and proteins, and copying "errors" are inevitable. The vast majority have little or no impact on the phenotypical expresssion of a living creature's appearance, attributes, etc., but the net impact within any animal or plant community is a fairly wide spread of potential characteristics. When nature "selects" a variation, it's not so much that a creature is "evolving to do something" (the teleological fallacy), but that a survival situation is occurring and those organisms favored by the genes they already happen to have suddenly have an advantage. Intelligence, however you define it, is a very useful attribute in a lot of situations, but one that requires a lot of physical apparatus to support. You should expect, in any wild population, a very considerable range of potential intelligence varying greatly from individual to individual. Sometimes, probably rarely, this will give a considerable reproductive advantage to those individuals who have a larger capacity for critical thinking. Other times, it will be irrelevant to their reproductive success, or be less important than other attributes.
*Intelligence
...is actually a very squishy concept. There is no objective measure for intelligence, and all attempts to create such a thing have been fraught with trouble. Ultimately, people are likely to mean more than one thing by the term depending on context. There isn't really an unbiased way of deciding whether an animal is on the whole more intelligent than another. The best you can do is compare how good animals are at performing certain specific tasks. Different cultures among humans, let alone different species, have wildly different ideas about what "good thinking" looks like, or which tasks are likely to be useful to perform. Western IQ tests, for instance, put an incredible amount of stock in one's ability to manipulate geometric shapes in your head, a talent rarely needed in real life, while considering social skills (like interpreting a complex facial expression for instance) irrelevant to intelligence despite the obvious and immediate utility of being able to predict someone's next action accurately. Intelligence is a deeply culturally bound notion, and IQ tests of any sort routinely have their adequacy challenged even when only humans are competing. Throw chimps and parrots and whales into the mix, and very quickly you're asking philosophical questions about what the mind is and how it can be assessed, leaving scientific objectivity a bit in the lurch.
*Parts
More than just anatomy is needed to support intelligence, but a lot of anatomy is needed to support intelligence:
-A complex neurological structure is needed to support certain kinds of reasoning. Size matters less than the number of connections, as many examples have shown. Octopi have tiny brains compared to humans, but are almost infamously clever problem-solvers in lab situations.
-The rest of the nervous system also needs to be quick and complicated to accomodate.
-You want a certain kind of brain; one that builds itself in interesting ways over the course of an organism's life, flexible and dynamic in its ability to make novel connections or use its basic structures in unique ways, rather than a brain that simply reproduces itself exactly from one generation to the next. Your connectome is more unique and singular than a fingerprint; a mayfly could not say the same.
-Sense organs need to be in such a state that they can deliver useful and meaningful information for the system to analyze.
-Longevity of life span matters a bit- you become more "intelligent" over time simply due to having more experience, and a lot of the animals we tend to think of as smart also have relatively long periods of infancy and adolescence.
-Culture, society, etc are also important factors, and a key difference between humans and other organisms. We learn from one another very easily, a rare trait among the living organisms and one that erases what is otherwise normally a limitation on intelligence. Intergenerational knowledge can accumulate nearly endlessly, instead of dying along with the first organism that produced it. Anthropologists are fond of saying that culture is the primary adaptive mechanism of our species, long ago replacing genetic mutations as the most important differences between one human group and the next. We have no idea when culture as such evolved, limiting our contemplation of the intelligence question in prehistory.
-The downside is that not everyone has equal access to the overall knowledge/skills of humanity, and social and political factors can severely limit any one individual's ability to reach their full natural potential of mental development.
* Human ancestors
We honestly have little access to the neurological anatomy of our forebears. No one can tell you with any certainty what an H. habilis or A. afarensis was like in terms of general intelligence, or how they would have done on an IQ test. There are hints - the fact that our crania, especially the frontal cortex, increased in size over time over the last two million years, and the rest of our body accommodated this; indeed, so quickly that our current giant heads are a major cause of deaths in childbirth, without this causing the expected effect in terms of reproductive success. Apparently it is better to be smart than to have successful births. There is also archaeological evidence; our tool-making capacity expanded over time, accelerating remarkably toward the end of the Pleistocene. You also see the first hints of religion and the like, and symbolic communication through letters or art, signs of what most would consider abstract reasoning. But this is nevertheless largely a blank spot, and there is a long history of would-be scientists coloring in those lines with modern agendas. The first IQ tests were, after all, devised largely for the purposes of justifying racial discrimination in schooling, even if that is not now seen as their primary function. Our biases tend to get very much in the way of analyzing questions of intelligence.
