I was going to comment on the papers from the book, ‘The origins of object knowledge’ on a paper-by-paper basis… but I’ve changed my mind. The editors have done a fine job and the papers interlink and cross-reference, so I’ll endeavour to summarise instead.
Every so often, I’ll be reading something and I’ll get an awful feeling – like the moment when Wiley Coyote has run off the edge of a cliff and suddenly realised that he is treading thin air – the feeling that a great intellectual chasm has just opened up beneath me. (Years ago it happened when I was trying to get my head into the fields of polycrystal plasticity and crystallographic texture, and realised that I probably wasn’t clever or mathematical enough.) I had that feeling again when reading this book. At least, with this topic, I think I understand what they’re arguing about. The editors of this book have used the differing papers and authors to reveal to the reader the current debates in the field of developmental psychology and object knowledge – I didn’t realise there would be such a debate, or such trenchant positions.
Anyway, for me the key points go something like this:
We encounter objects from the moment we are born. Based on his observations Piaget made a series of statements about the developmental stages of infants and how they understand the world around them. More recent research has suggested that infants develop much more sophisticated ‘intuitions’ about objects much earlier than Piaget proposed.
The basis for these new findings is an elegant experiment. It is based on the notion that if an infant is surprised by something, i.e. if it does not behave in the way that the infant expects, then the infant will stare at it for longer than they would stare at something that behaves ‘normally’. From this researchers have conducted various experiments, whilst observing looking time, to see what phenomena evoke an infants interest and, from there, to deduce the ‘intuitions’ that the children have. I’m using the word ‘intuitions’ here because there is some question about whether it can be counted as knowledge, in that the expectations are based on an implicit rather than explicit framework of understanding.
So, from as early as 2 ½ months, infants have an intuitive framework of expectations about this class of things that we would call ‘objects’. These expectations revolve around object properties and behaviour – that solid objects will not pass through one another; that inert objects will continue travelling in the same direction, that solid objects retain their form. They also have expectations around number, that a number of objects will remain consistent and have a sense when that number is changed.
Initially, the key property that children use to identify objects is their shape or form. This is domain dependent – the initial, key property of food is colour. As children develop their means of categorising objects becomes more complex. Nonetheless, the catastrophic failure of an infant’s working memory, when its capacity for objects is exceeded, is mollified when the objects being observed are more complex. This suggests that while shape is the primary factor, other features such as colour and texture do matter.
Children’s brains, particularly the frontal cortex and parietal lobes continue to mature up to 5 years of age. It is known that the visual systems in the cortex are deeply modular so it is possible (I’m speculating) that the modularity develops during this maturation period. Even if it did, this seems like a chicken and egg problem: we know that the brain is plastic and develops in response to input and usage (e.g. the brains of taxi drivers), so it could be that infants’ increased utilisation of the richness of their environment stimulates the modularity, or it could be that the increasing modularity facilitates the greater complexity of children’s intuitions.
Infants are also able to respond to an object’s apparent agency. That is their expectations of object behaviour will be modified depending on whether they can perceive that the object is inert, moved by an external agent, or has some internal motive force (i.e. a mechanical toy or animal).
One of the key debates in this area is whether or not these expectations are hard wired into the brain or whether they emerge out of their perceptions of the world around them in the first couple of months. One group argues that this set of ‘core knowledge’ is innate, another argues that it is learnt. There are unresolved issues in the arguments of both camps. And, for me, it is sufficient to point out that some work using connectionist (neural net) modelling has shown that the early emergence of these expectations can be explained without necessarily invoking the notion of core knowledge.
As I mentioned earlier, the notion of ‘knowledge’ is problematic both because the children cannot voice these intuitions and because (results show that) they cannot act on them. Piaget showed that infants consistently incorrectly reach for an occluded object and made certain deductions from that. More recent similar research has shown that, when objects are hidden, young children look at the correct location but reach for the wrong location. This dissociation between looking and reaching perseveres until as late as 2 ½ years of age. The reasons behind this dissociation are, currently, not clear. It could be caused by children relying on earlier (mistaken) behaviours to guide current behaviour, or it could be caused by the different rates of development of the different neural components (ventral and dorsal) that are thought to guide reaching behaviour, or it could be a combination of things. Certainly the computational modelling presented in the book indicates that differing rates of learning could reasonably contribute to this dissociation.
At the heart of this debate about object knowledge is, of course, a debate about object learning. One paper (Xu et al.) looks at this more closely. They suggest that the nativist (i.e. core knowledge) model does not provide a mechanism for learning, whilst the empiricist model of associative learning is too slow to explain the rate of children’s development. Xu et al.argue for another model of learning, one of rational constructivism. “The rational constructivist view is committed to [the idea that children’s intuitions are] learned, are subject to domain effects and to be learned by domain general inferential mechanisms such as overhypothesis formation.” Their argument and research is very persuasive.
This area of the nature of learning is an avenue that needs pursuing further. Thankfully, there are some review papers out there that I have downloaded. Also, these papers deal with very young infants (generally from 2 months to 2 years), so I still haven’t pinned down where primary aged children are at when they encounter objects – cortical maturation may be over (for a time) but their learning certainly isn’t.