The authors wanted to examine, through causal and noncausal events, if perception of cause and effect develops over time or is part of our innate endownment. They set about doing this from the launchpad of Michotte's (1963) pioneering work with adults, and subsequent investigations of infant causal perception (Leslie, 1982; Oakes and Cohen, 1990).Different events have been produced as animation sequences and presented to infants. 1) A causal direct launching event in which one object moves and contacts a second object, which instantly moves away. 2) A delayed launching event where the second object only moves after a temporal delay following impact.

3) A delayed launching without collision, which involve a delay and lack of contact.Habituation of looking procedures has been used so the infant is habituated to a specific type of event, and then is presented with a different type during testing. If infants attribute a special causal status to direct launching, they should dishabituate more if the test event differs from the habituation in terms of causality compared to if it does not.A key experiment by Leslie (1984) put forward the argument that infants' causal perception is much like adults, and subject to little or no developmental change.

However Cohen and Amsel (1998), with support from previous research Cohen (1998) and Oakes and Cohen (1994), have argued on the basis that causal perception is the result of systematic development through the first year. The two approaches to the question of causal development in infants represent a modular, nativist framework, predicting similar results in all three groups, in this article referenced through Leslie's work (1982; 1984), and a constructivist developmental framework. This predicts recovery of attention to events differing in terms of causality only at older ages, with recovery at younger ages based on simpler, perceptual distinctions between the events, (Cohen and Amsel, 1998, p717). They suggested that continuous movement play an important role in very young infants' processing of launching events.

Cohen and Amsel supported their hypothesis by showing that before 6 months of age infants could not process causality in a direct launching event, demonstrating that the processing of causality is certainly not inevitable during infancy and remains sensitive to specific characteristics at particular ages. It was most likely to develop at 6 months and develop from there. They found that the 6 1/4 month old infants who had habituated to a noncausal event dishabituated more to a direct launching than to a noncausal event during trials.The concept reflected in the term precursor is narrower than the concept that is reflected in the term cause. A precursor is an event that precedes and indicated the approach of another event. The extent to which the precursor is prerequisite varies with the theory suggested.

The evidence necessary to support the suggestion that some ability is a precursor is indirect, as it is for causal factors, no experimental design can establish that any ability is a precursor of another ability.The introduction gives a clear guided path into the area of the subject. Immediately it sets out the use of Michotte's (1963) launching events, as this is a major part of the design for all the experiments, it is important for it to be set out early. Also within the first paragraph it sets out the theme with which the article will expand upon, effectively a nature-nurture debate. Though it would be beneficial if the authors explained that they are two dimensions and not two mutually incompatible worlds.

However, the second paragraph seems to deviate slightly from the rest of the article, as it describes earlier research in the field employing occluded screens. It would be more beneficial for the pacing of the essay to cut this part out and replace it with more relevant research and a review of Leslie's (1984) experiment, so as to get an understanding of the different arguments earlier on. Belanger and Desrochers (2001) incorporate just such a process in a similar article, although the review of the literature is too brief to fit in as an alternative viewpoint to the author's logic.In terms of method the experiment's primary goal was to replicate Leslie's results and as such the experiment was designed to be as similar to Leslie's original as possible. The experiments will in addition go on to study progressively younger infants so that the methodology is justified and, therefore, appropriate to answer the questions identified. However, as Leslie's experiment had results that infants at 6 1/4 months respond to Michottian event sequences in terms of causality, the current authors could have expanded on this and looked at ways to find the extent to which infants of this age could perceive causality.

Examples would be the use of more complex stimuli, also reaction to entraining events, used by Belanger and Desrochers (2001).As the authors suggested that continuous movement play an important role in very young infants' processing of launching events an important parameter thought to affect their visual tracking is stimulus velocity. This concerns the influence of velocity on infant attention, Nelson and Horowitz (1987); in general younger infants are less sensitive to slower motions than older infants are. So rather than processing it as one continuous smooth motion as suggested by Cohen and Amsel, it could be that the infants prefer to look due to the speed at which it passes from one end of the screen to the other.

A criticism, which can be levelled at the procedure incorporated into the experiment, could be the amount of time the infants had to look at the stimuli for it to begin or end. The experimenter would begin the trial as soon as the infant was looking at the attention-getter; also the event sequence would end if the infant looked away for 5 seconds. Later research in similar studies, however, would initiate the trial when the infant looked at the stimuli for at least 0.5 seconds (Belanger and Desrochers (2001) or 2 seconds (Cashon and Cohen 2000) and had to look away for a minimum of 2 seconds to end the trial. Such a procedure would help in controlling fatigue events. More importantly as the original procedure was designed solely for 6 1/4 month old infants, the attention duration for the initiation and ending of the trials should be such so as to be of benefit to all the age groups.

The results found can only be adequately integrated within a developmental framework. The processing of causality would only begin at around 6 months. A similar developmental sequence to what empirical studies of Piaget's (1954) sensorimotor causality revealed (Desrochers, Ricard, and Gouin Decarie, 1995).Cohen and Amsel's conclusion follows on logically from their reported results they expand on the theory that it is possible that continuous movement plays an important role in very young infants processing of launching events. Although they only speculate about the differences, this in turn gives the opportunity for additional future research to confirm their theory.Finally, as much of the article concerned the nativist framework expounded by Leslie, and the results of Cohen and Amsel supported their own constructivist framework it would have been helpful if the article could suggest why Leslie's theory was not supported.

This could be due to its broadness. A could be accurate if it was more specific. As suggested by Belanger and Desrochers (2001), a limited module which processes causality in a direct launching event. So keeping it simple and identifiable, applicable only to specific classes of launching events at given ages.