We know that babies begin learning at birth -- if not before, in the womb, becoming familiar with their mother's voice.  But we also know that autobiographical memory (ABM) -- the ability to remember specific events and experiences -- develops much later.  Why this is so has been addressed, to some degree, in the theories of infantile and childhood amnesia described above.  Perhaps the most comprehensive account of the development of  ABM has been provided by Robyn Fivush (see, for example, her article with Nelson in Psychological Review, 2004; and her chapter in the Annual Review of Psychology for 2011, from which this account is mostly drawn).

Fivush first addresses the question of whether ABM is unique to humans.  Following Tulving (2002), she identifies two components to ABM:

• Episodic memory, "memory of the specific what, where, and when of an experience;
• Autonoetic consciousness, a sort of "mental time travel" entailing "the awareness of self having experienced the event n the past".

Fivush argues that while specifically episodic memory may be available to some nonhuman species, ABM includes more.

• Memory of the self as the experiencer of the event (I would add that the self might also be remembered as the agent, the patient, or the stimulus).
• Chronological organization of a life narrative including the past, present, and future.
• The use of memory to serve social functions such as self-definition, relating to others, and self-regulation.

Fivush, then, defines ABM as a form of cultural activity which is shaped by social interaction.  But it is also a developmental accomplishment requiring a number of component skills, including:

• subjective consciousness;
• the ability to link one's past and present selves;
• the ability to construct a personal time.

Fivush begins with a fairly traditional taxonomy of memory, adopting Squire's distinction between declarative and nondeclarative (i.e., memory, and Tulving's distinction between episodic and semantic memory.

• Recall that Squire's category of "nondeclarative" memory includes procedural knowledge, conditioning, and priming.  Squire's category of "declarative" memory, then includes both explicit episodic memory and semantic memory. 
  
• Infants may show various forms of "nondeclarative memory, such as procedural learning and priming, before they show declarative memory. A good example of this is Rovee-Collier's work on memory in infants.
• Fivush makes the distinction noted above between episodic memory and ABM, based on autonoetic consciousness.
• Again, Rovee-Collier's work seems to show that infants have the ability to remember episodes -- though one could argue that her learning setup is simply a variant on instrumental conditioning, and shows only that infants can acquire and retain conditioned responses.
• More convincing evidence in this regard comes from work by Bauer (Current Directions in Psychological Science, 2007) showing that 9-month-olds can remember a specific experience which happened to them only once several weeks before.  That's a lot closer to episodic memory.

She then argues that ABM requires both episodic memory and autonoetic consciousness.  And in order to achieve autonoetic consciousness, you've got to have a self (which gives you the "auto" in "autonoetic").  It's that subjective sense of selfhood that develops over time, and which allows ABM to develop over time as well.  She further argues that "it is not possible to create an autobiography outside of a social-cultural group" (p. 563).  Which gives rise to her Socio-Cultural Developmental Theory of ABM, which she began developing along with Katherine Nelson.

Socio-Cultural Developmental Theory

Fivush argues that while a capacity for episodic memory is universal within the human species, "the forms and functions of autobiographical memory are socially and culturally variable".  The nature of ABM will depend on assumptions about the self -- and, for that matter, memory -- which characterize the social-cultural context in which the child grows up.  The result is that both the content and the functions of ABM are likely to vary from culture to culture -- and especially between Western and non-Western, and between literate and non-literate, cultures.  As she points, out, in Western culture, at least, "Being a person means having a story to tell about your experiences, and these stories help constitute who you are".

• Language is critical to this process, because it is by means of language that we share our past with others, and engage in joint reminiscence.
• And so is the construction of narratives -- not just the chronological organization of events over time, but also causal linkages between them, comparisons between events, and identification of their significance for the individual.

With those points in hand, Fivush then goes on to trace the emergence of autobiographical memory.

• The fetus in utero has some capacity for learning and memory, as reflected in the ability of neonates to distinguish between their mother's voice and novel sounds.
• In infancy, the ability to distinguish between familiar and novel sounds is apparent by 6 months.
• Deferred imitation, which obviously depends on memory, is established by 9 months.
• By 12 months, infants can reproduce complex sequences of actions that they had observed only once, some weeks before.
• The development of language also creates the possibility of verbal recall of past experiences. 
  
• By 16-18 months of age, a point at which normal infants are able to string two words together, they are spontaneously referring to past experiences such as eating Cheerios or going to the park.
• Adults respond by elaborating on these verbalizations (e.g., "Go park") with more coherent narratives (e.g., about going to the park and what went on there).
• In this way, adults begin to model narrative storytelling for children.
• Of course, some adults don't do this, and some adults do it better, or at least differently, than others.
• "Elaborative" mothers ask open-ended questions which encourage the child to recall more information.
• They also integrate their child's responses into a coherent narrative
• And they also provide hints about aspects of the event that the child has not yet recalled.
• "Maternal reminiscing style" is highly correlated with the development of ABM in children.
• These correlations have been confirmed in experimental studies in which mothers have been trained to adopt a more elaborative style.
• Mothers are more elaborative with their daughters than with their sons -- thus modeling what becomes an adult gender difference in elaborative storytelling.
• And elaborative style is more common in Western ("independent") as opposed to Eastern ("interdependent").
• At the same time, the child is developing a sense of him- or herself as a self -- as an individuated person.
• First and foremost, this entails the development of subjective consciousness -- in particular, that the remembering person is recalling representations of his or her own past experiences.
• The child must also be able to link his past with his present (and, for that matter, future) self.
• And the child must be able to establish a personal timeline, or chronological sequence of remembered events.
  

The result of all this is that, by the end of the pre-school years, children have ABM -- the ability to tell coherent stories about their past experiences.  It is this social-cultural developmental process, rather than the development of the hippocampus, which accounts for the end of infantile and childhood amnesia.

Episodic Memory  and Consciousness in Animals Fivush argues that ABM is unique to humans, but she concedes that episodic memory may be part of the cognitive repertoire of nonhuman species.  In fact, a number of studies have shown that some non-humans do have the capacity to remember specific events that occurred at specific times, and can even plan for the future.   For a review of this literature, see the article by Clayton et al. in Nature Reviews Neuroscience (2003). So where does the claim that ABM is uniquely human come from?  Apparently, Fivush, like Tulving, believes that autonoetic consciousness is uniquely human.  Perhaps: without the ability to interrogate animals about what they're thinking, we may never know.  But then again, the whole point of research in animal cognition is, precisely, to find out what animals are thinking.  Given the Darwinian principle of evolutionary continuity, it is probably not correct to deny that all nonhuman animals have autonoetic consciousness.  And if they do, then at least some species may have the capacity for ABM as Fivush defines it.