Open Book Publishers logo Open Access logo
  • button
  • button
  • button
GO TO...
book cover

4. Discontinuities in Development1

© 2017 Patrick Bateson, CC BY-NC-ND 4.0

Many aspects of body and behaviour change markedly during the course of development, sometimes quite suddenly. Discontinuous change is most obvious during the first two decades of a human life, for example at birth and puberty. Such discontinuities are not mysteries. Many physical and biological systems are capable of changing in an abrupt, discontinuous way. Steadily increasing the pressure on a light switch does not produce a steady increase in the brightness of the bulb it controls. The switch has a point of instability, so that one moment the bulb is dark and the next moment it is fully lit. Similarly, a relatively small internal or external change can quickly transform a developing organism’s characteristics to something that looks quite different. The fertilised egg of an animal rapidly divides becoming a ball of cells, the blastula. The cells continue to divide, but do so at slightly different rates. The steady change is such that the blastula suddenly seems to collapse on one side like a deflated rubber ball and a two-layered structure called the gastrula is formed. The embryo has changed its appearance dramatically as a result of a process of continuous growth.

Sudden changes in behaviour during an individual’s development may have biological utility, reflecting the changing ecology and needs of the individual as it gets older. The relatively abrupt alteration in the method of feeding at weaning, or in the mode of behaviour towards members of the opposite sex at puberty, are obvious enough. Becoming an adult involves a long process, and at each stage in development the individual must cope with challenges of the particular world in which it is living at the time. The caterpillar feeds voraciously and has adaptations for doing so, and by its camouflage, warning coloration or behaviour avoids as best it can the attentions of other animals that would eat it. Then its body goes through a dramatic change while a pupa and it emerges as a butterfly. Such metamorphoses have no exact parallel in humans but the requirements of living in a very different world from that of an adult are real. Are the apparent discontinuities seen in development related to the changes in function?

Monarch butterfly before and after metamorphosis. Monarch butterfly caterpillar. Photo by Antilived (2006), Wikimedia,, CC BY 3.0. Monarch butterfly. Photo by LyWashu (2008), Wikimedia,, CC BY 3.0.

In the behavioural literature, ‘discontinuity’ is used in several different senses. Strictly, it should refer to a sudden alteration in the characteristics of a system. So, for instance, when kittens suddenly start to take solid food in appreciable quantities, their rate of weight increase shows a marked rise since, before it happened, the mother was no longer able to provide sufficient milk to sustain the needs of her growing offspring.2 ‘Discontinuity’ is also used for the reorganization of rank on a given measure, so that when scores of attentiveness, vocalization, and sniffing in two-month-old human infants fail to predict their scores on the same measure at four months, a discontinuity is said to have occurred. It is obvious that a discontinuity in the first sense of a change in organisation does not necessarily imply a discontinuity in the second sense of the rank ordering of individuals. The rank order of the kittens in terms of weight could remain unchanged with one kitten remaining the heaviest throughout development. In fact, the sudden change occurs at different ages in different litters, so the rank order may be altered, but only temporarily until some time after all the kittens have started to eat solid food.

Another type of evidence for discontinuities comes from longitudinal correlation studies in which the same pattern of behaviour is measured repeatedly in a group of subjects through development. Behaviour at one stage may successfully predict behaviour at the next and the next but not the one after that. So long as the rank order is dependent on new sources of variation or fluctuates randomly, extreme values tend to regress toward the mean. As a result, correlations between behaviour at one stage and behaviour at succeeding stages tend to diminish with the lapse of time. Even so, a particularly rapid reduction in the strength of a correlation requires additional explanation.

Associations between different measures side-step the problem of whether new behaviour patterns really are new. But it creates a fresh problem, which is whether the ‘old’ behaviour patterns really are the same when measured at later stages of development. The superficial descriptive similarities between different behaviour patterns may be deceptive. A mistake of this type might be especially likely when the measure in a standard test was something like ‘Latency to approach’ or ‘Time near object’. Careful observation of the descriptive structure of the motor patterns might indicate that quite different systems of behaviour are involved at the different stages of development. For instance, when domestic chicks approach a novel visually conspicuous object, they may be responding socially at one day old but asocially exploring it when they are one week old. The character of the discontinuity can be investigated by examining not only the detailed form of the behaviour but also the consequences for the animal of its response before and after the change.

Changes in rank order need to be handled with particular care because the changes can usually be explained in a variety of ways, some of which are trivial. Positive correlations between a measure taken at one stage of development and those taken at others may evaporate for an uncomplicated reason, namely, that the measurements start to clump together. This clumping might arise because the measurements reach some ceiling (or floor) on a scale of measurement. For instance, a correlation between the performances of children on a simple test of addition at one age and their performances on the same test at later ages might drop to zero because a point is reached when all the children get full marks.

An absence of a correlation between two measures as determined by a straightforward statistical test need not necessarily mean that no relationship exists. A change from a linear to an inverted U-shaped relationship might easily be misinterpreted as a loss of any relationship. Consider a case in which IQ is repeatedly measured at different times in a group of children. Suppose that the test is one that is especially appropriate for a certain age range and contains questions like ‘Continue the series O,T,T,F,F…’ Bright children have no difficulty in providing the answer: ‘S,S,E…’ Bright adults rack their brains looking for complex rules that might provide some meaningful sequence and totally miss the point that the series consists of the first letters of One, Two, Three, Four, Five… Children who do well on an IQ test at one stage switch to a more adult-like problem-solving strategy before their peers and consequently perform relatively badly on what is for them an inappropriate test. So an inverted U-shaped relationship between cognitive ability at the two stages might easily arise. The change in the children’s strategy would be interesting but would not signify that a fundamental reordering of the children had occurred as a result of some event in development. Clearly, this possibility can be examined by plotting the data in graphical form. Furthermore, with appropriate measures of behaviour at later stages in development, it should be possible to show that the original rank order is eventually regained.

In a cohort of individuals growing older together, their rank order on a particular measure at Stage 1 might correspond closely to their rank order at Stage 3. For instance, the factors responsible for the difference in their height at Stage 1 might be the same set of factors responsible for the difference in their height at Stage 3. A quite different set of factors might be responsible for the timing of a growth spurt. Consequently, the rank ordering of height at Stage 2 during the period of the growth spurt might be quite different from that of Stage 1 and Stage 3. Such an effect should be detectable from the data, provided of course that the cohort has been sampled at appropriate points during development.

Changes in development take many forms, which invite different explanations. A qualitatively new pattern of behaviour may appear and an old one may disappear; switching from suckling to eating solid food during weaning is an example. An individual child’s consistent tendency to cry more than it laughs might change at a particular age. The rank-order of individuals at one age often fails to predict the order at another age, so an infant who is, say, more attentive than others at two months may be less attentive than others at four months. Sometimes, the slow starter catches up and overtakes the more precocious child. Parents who are proud that their child is reading at the age of four should not assume that the child will turn out to be a genius. Faster development does not necessarily mean a superior outcome.

Loss of continuity

Continuity from one age to the next may be lost for many reasons. One is that development is affected by many influences, not all of which are the same for everybody. Another is that children are often profoundly influenced by the social situation in which they find themselves.

Continuities across age may also be lost temporarily because different children pass through a particular transition at different chronological ages. People who are tall for their age when they are two years old are also highly likely to be tall when they are twenty. But they may not be tall for their age at thirteen, because individuals differ in the age at which they undergo the growth spurt before puberty. In this case, the property of being taller than peers survives the big changes occurring at puberty. The same is true for many distinctive aspects of behaviour and personality. When these change permanently, as undoubtedly they sometimes do, it may not be because the person has passed through one of the supposed metamorphoses of development.

A change in control of an unchanged motor pattern might occur at a certain stage in development. Three-week-old kittens with their eyes recently open approach the mother from a distance for suckling but when they reach her, they search for a nipple with their eyes closed.3 Gradually, vision comes to trigger not only the approach to the mother but also the nipple-searching behaviour patterns that were formerly triggered by non-visual sensory systems. Such a change in control might easily be associated with a change in the rank order on the scale of measurement of the behaviour. In this case, the discontinuity can be investigated by an examination of the conditions that elicit the behaviour.

Mother cat suckling her kittens. Photo by Ashim 71 (2014), Wikimedia,, CC BY 4.0.

Every human individual must act many roles during his or her lifetime. Leaving aside the accuracy of how the roles are defined, how much of the individual’s personality and distinctive behavioural characteristics fails to survive the crossing of the boundary between one of the distinct ages and the next? A big change occurs in humans between the ages of two and four with the emergence of language and an awareness of self. Few adults remember much of what happened to them in their first few years. Even if they are subjectively certain that they remember their birth, the corroboration is invariably suspect or missing. It might be argued that in the first few years children have no memories; nothing has been stored so nothing has to be erased. Such a view is clearly false. Young children have good functional long-term memories. In one experiment, for example, children around two years of age were asked to imitate actions that they had seen eight months before. They performed significantly better than children who had not previously seen these actions.4 The absence of memories from infancy does not reflect an inability to form enduring memories at the time, suggesting instead that young children can remember things but that substantial reorganisation of memory occurs between the ages of two and four.


The phenomena of discontinuities in development excite considerable interest because the changes may represent an alteration in the ecology of the developing individual. A variety of different explanations may be offered in any particular case. The phenomena are undoubtedly heterogeneous and so, while each explanation could apply some of the time, it is doubtful whether any one of them will apply all of the time. Some of the explanations point to deficiencies in method; some imply that after a loss of the original rank order on a particular measure, that order will be recovered, and some suggest that a change in rank order is permanent. The analogy with metamorphosis would be relevant only to phenomena brought about by the last group of processes. When behaviour at a given stage of development is matched to the ecological conditions in which the individual lives, the appearance of design is raised once again.

1 This chapter is drawn in part, with permission of Springer, from Bateson, P. (1978), How does behavior develop? In: P. Bateson & P.H. Klopfer (eds.), Perspectives in Ethology. Vol. 3: Social Behavior. New York: Springer, pp. 55–66. Some animal examples are given but the chapter focuses particularly on discontinuities in human development.

2 Martin, P. & Bateson, P. (2007), Measuring Behaviour. 3rd ed. Cambridge: Cambridge University Press.

3 Bateson, P. (2014), Behavioural development in the cat. In: Turner, D.C. & Bateson, P. (eds.), The Domestic Cat: The Biology of its Behaviour. Cambridge: Cambridge University Press, pp. 11–26.

4 Strohl, K.P. & Thomas, A.J. (1997), Neonatal conditioning for adult respiratory behaviour. Respir. Physiol. 110.2–3, 269–275,