Critical Periods (Sensitive Periods) in Puppies – Revisited

2015 could be declared an import year for animals on several different fronts. One reason might be that in April, a New York judge made legal history by declaring that 2 chimpanzees were to be recognised as ‘Legal Persons’ – so-called habeas corpus (Bekoff, 2015).

PHOTO: courtesy of Caitlin Hume, Springer Rescue for Scotland. A charity that successfully rehomes over 150 Spaniels every year.

Following this breakthrough for animal welfare, the New Zealand Government passed the ‘Animal Welfare Amendment Bill’ that declared that animals are sentient beings – that is, they share some of the same basic human emotions such as happiness, sadness, grief etc. (McIntyre, 2015).

Another reason might be that 2015 marks the 50th anniversary of one of the most important dog books ever written, the ‘Genetics and the Social Behavior of the Dog’ by John Paul Scott, and John Fuller, published in 1965 (Scott and Fuller, 1965). This book has sold millions of copies and is still in print and widely available today for just a few pounds. Known as ‘ Scott and Fuller’ for short, it is so important that all good dog breeders know about it and apply the wisdom contained in it to raise their litters of puppies, even though they may never have read it. Animal trainers and behaviourists know about it because it helps them run good puppy classes and explains why some behaviour problems occur in dogs. The Kennel Club / Dogs Trust recently launched version 2 of the ‘Puppy Plan’, written by (Menteith, 2014) is based on it as is every book, article and seminar about raising puppies. The knowledge that Scott and Fuller gave the world is that all puppies have a well-defined neonatal period, a transition period and a critical (or sensitive) period, and what happens to them during these periods shapes what kinds of dogs they grow up to be.

Scott and Fuller carried out their pioneering work at the Bar Harbor research facility where they meticulously catalogued everything they did. The significance of this work is that it was carried out in a controlled, laboratory environment in which each generation of dogs could be raised in a uniform way and be subject to the same, controlled early experiences. Scott and Fuller conducted extensive research over a period of 14 years on successive generations of 6 breeds of dog (Basenji, Beagle, Cocker Spaniel, Shetland Sheepdog and Fox Terrier) and their hybrids in an attempt to quantify breed differences in behaviour. What they found was that the critical period starts at around 3 weeks of age and ends at around 12 weeks of age. During this time, the puppies would readily approach novel objects. They would also readily approach other animals and humans to which they would become ‘socialised’ if given the opportunity to do so. Scott and Fuller also noticed that puppies developed a ‘startle response’ at around 20 days of age and that a fear response also emerged sometime during the critical period, and this varied between breeds – more about this later.

As well as the landmark periods of development, Scott and Fuller’s research also showed some clear differences between the breeds they studied. For example, the Terriers, Basenjis and Beagles were more emotionally reactive than Cocker Spaniels and Sheepdogs. Cocker Spaniels were the most emotionally stable, the easiest to train and the most obedient, while the Basenjis were the least co-operative. The hunting breeds, Basenjis, Beagles, Cocker Spaniels and terriers, did better than the Sheepdogs in problem solving skills.

Following on from Scott and Fuller, 2015 marks the publication of another important study in the Journal of Veterinary Behavior (Morrow et al., 2015). This study focuses on the age at which the fear response emerges (i.e. the onset of hazard avoidance, see the discussion for more details) in 3 different breeds of dog raised in their own homes by their breeders. Although necessarily less controlled than the Bar Harbor work, this study arguably gives us more information about puppies destined as pets because they are reared in more ‘natural’ environments. The main reason why this study is important is that an understanding of how and when puppies develop the emotion of fear, and how it is dealt with, is crucial in understanding how to prevent this leading to potentially life-long behaviour problems as adult dogs.


How the research was done
In contrast to the Bar Harbor research, the researchers decided to recruit dog breeders and travel to their homes to study the puppies as they grew up on-site. This study design inevitably imposed limitations on the scope of the study, specifically – the availability of enough puppies within a breed, the willingness of the breeders to participate and cooperate, the logistics and costs of travel to see the litters on multiple occasions through the study period and the potential of other, unforeseen variables arising that could skew the results. Nonetheless, such a design has one big advantage – studying puppies in their own homes as they prepare to go out into the big, wide world and join their human families eliminates any influences that might be imposed on their development in the more controlled, sterile environment of the laboratory.

Within these limitations, the researchers managed to find enough puppies across 3 different breeds to allow them to collect enough data for analysis. The breeds chosen where the German Shepherd, the Yorkshire Terrier and the Cavalier King Charles Spaniel. These breeds were good choices because they represent 3, distinctly different types of dog in terms of temperament and physical appearance – a herding / guarding breed, a working terrier breed and a family pet, ‘luxury’ breed respectively.

In terms of numbers, 98 puppies in all were recruited, namely 5 litters of German Shepherds (33 puppies) from 3 breeders, 9 litters of Yorkshire Terriers (32 puppies) from 3 breeders and 7 litters of Cavalier King Charles Spaniels (33 puppies) from 5 breeders.

The puppies were visited and tested once a week for 6 consecutive weeks, starting at 4 weeks old and finishing at 10 weeks old. Because some of the puppies left for their new homes before they reached 10 weeks old, there were a few missing data points by the end of the study. In addition, it was found that at 4 weeks old, most of the Cavalier King Charles Spaniels were too immature to be able to collect meaningful data from them, so testing in this breed started at 5 weeks old instead.

Having recruited the 98 puppies, the first thing the researches did was to divide each breed group into 2 equal-sized groups – one group was to be the ‘TEST’ group and the other group was to be an untested ‘CONTROL’ group. The allocation-to-group was randomly done to ensure that there was an unbiased spread of dogs within each breed group across the respective breeders.

The TEST group puppies underwent 4 tests at each of their weekly visits. The CONTROL group puppies where placed on the test apparatus in exactly the same way and for the same length of time as was done with the TEST puppies, but they did not actually undergo any of the tests themselves. Doing this ensured that any influence that the test apparatus itself may have on the puppies behaviours was accounted for in the data collected from both the TEST and CONTROL groups.

The 4 tests were:-

  1. Novel Item Test: the novel item was a battery-operated toy duck that, when switched on, lit up and quacked as it moved forward. It was placed onto the test equipment with the puppy and in front of him, left for 5 seconds, after which the puppy was removed. The short duration of exposure was to minimise the effects of habituation during successive exposures. The test equipment was a black board marked up with white painted boundaries, like a football pitch, in order to give it a front, central and back area. The toy was placed at the front position and the puppy in the middle area. It was then easy to measure the puppies reactions as either an ‘approach’ (take a few steps across the line into forward section), an ‘avoid’ (take a few steps across the line into back section), or ‘neutral’ (stay within the boundaries of the middle section). The CONTROL puppies were simply placed onto the equipment for 5 seconds, their reactions recorded without being exposed to the toy duck, then removed.
  2. Seesaw Test: the seesaw was similar to the novel item equipment, but balanced onto a 10 cm high fulcrum. Each puppy was placed into a square painted on the central area of the board, left there for 10 seconds then removed. The puppies reactions were recorded as either an ‘approach’ (moved around the middle section of the board as if to explore), a ‘double-approach’ (moved to the edge and/or stepped off the board as if to explore with more enthusiasm), or neutral (stayed in the position placed). CONTROL puppies were placed onto the same equipment, but without the fulcrum.
  3. Step Test: the step in this test was a small platform that varied in size and height depending on the size of the puppies. The Yorkshire Terriers and the Cavalier King Charles Spaniels had a smaller and lower step 5 cm high, while the German Shepherds had a larger and higher step 10 cm high. Each puppy was placed onto a marked central point of the step, left there for 10 seconds then removed. The puppies reactions were recorded as either an ‘approach’ (moved off the step as if to explore), or neutral (stayed on the step). CONTROL puppies were placed onto identical equipment, but flat with no step.
  4. Loud Noise Test: The test equipment was the same as for the novel item test – a black board marked up with white painted boundaries. Each puppy was placed in the central area of the board, then within 3 seconds of being placed in position, the small step used in the step test was dropped about 41 cm in front of him from a height of about 51 cm. This generated a noise of about 95 dB (by comparison, an average dog bark is about 80 dB and a kitchen blender is about 100 dB). The puppies reactions were recorded for 5 seconds either as either an ‘approach’ (take a few steps across the line into forward section), an ‘avoid’ (take a few steps across the line into back section), or ‘neutral’ (stay within the boundaries of the middle section). CONTROL puppies were placed onto the equipment for 7 seconds without dropping the step test, then removed.

To ensure consistency, the same person performed all of the tests and the tests were undertaken in the same room at the breeder’s house away from the litter’s nesting box. Puppies were taken to the test room, tested, then returned directly to their litter afterwards. Each puppy was presented with all 4 tests, one after the other, on each visit. The order in which each puppy was tested was randomly determined on each visit. The order that the tests were done on each visit was also randomly allocated. This crucial step in the design of the study helped rule out an ‘order effect’ that might influence the data being collected.

Everything was meticulously recorded using a digital video camera to ensure that all the puppies reactions to the proceedings could be analysed and interpreted by the researchers. In addition, the recordings were also analysed by 2 other independent and knowledgeable individuals that were not present at the times of testing and did not know what interpretation the researchers had made. This ensured that, as far as was possible, what was being noted down as data was consistent and valid.

In addition to the 4 behavioural tests, the researchers also collected 2 saliva samples from each puppy using specialised salivary swabs. This was only possible from the Cavalier King Charles Spaniels and the German Shepherds, however because the Yorkshire Terriers were too small to make this practical. The first sample was taken before the daily tests were carried and the second 20 minutes after testing was complete. The saliva samples were analysed for their cortisol levels which is a physiological marker for arousal, or stress.


The purpose of this study was to see if the fear response consistently and reliably emerged at different ages in the 3 different breeds of dogs studied. A fear response was considered to be present if a puppy showed for the first time a quick and immediate avoidance behaviour to any one of the 4 tests on one of the weekly test days. In other words, the puppy was now perceiving that test as a hazard, and reacting accordingly with a fear-related avoidance behaviour, compared to previous weeks when the puppy approached or ignored the test.

The proportion of dogs in each of the TEST groups that showed the emergence of hazard avoidance behaviour during the 6 week trial varied:-

  • Cavalier King Charles Spaniels: 53%.
  • German Shepherds: 28%.
  • Yorkshire Terriers: 78%.

The average onset of hazard avoidance in the 3 breeds of dogs were:-

  • Cavalier King Charles Spaniels: 54.8 ± 2.74 days.
  • German Shepherds: 39.4 ± 1.60 days.
  • Yorkshire Terriers: 43.6 ± 2.48 days.

The onset of hazard avoidance behaviour emerged later in Cavalier King Charles Spaniels than in the other 2 breeds, and hazard avoidance behaviour in German Shepherds and Yorkshire Terriers emerged at the same time statistically.
Neither the gender of the puppies, nor their breeder location had an effect on the results.


Analysis of the puppies reactions to individual tests also showed significant breed differences:-

  • Novel Item Test: Up to the age of 6 weeks, there were no differences. At 7 weeks the German Shepherds and the Yorkshire Terriers in both the TEST and CONTROL groups showed more ‘approach’ behaviours than did the Cavalier King Charles Spaniels. At 8 weeks the German Shepherds had overtaken the Yorkshire Terriers and showed more ‘approach’ behaviours.
  • Seesaw Test: Up to the age of 6 weeks, there were no differences. At 7 weeks the German Shepherds and the Yorkshire Terriers showed more ‘approach’ behaviours than did the Cavalier King Charles Spaniels. At 8 weeks the German Shepherds had overtaken the Yorkshire Terriers and showed more ‘approach’ behaviours.
  • Step Test: There were no age-related differences in the behaviours of any of the puppies observed in this test. However, both the TEST and CONTROL groups of the German Shepherds and the Yorkshire Terriers showed more ‘approach’ behaviours (moved off the step as if to explore, or for the CONTROLS, moved off their placement positions) than the Cavalier King Charles Spaniels. The CONTROL groups of the German Shepherds and the Cavalier King Charles Spaniels showed more ‘approach’ behaviours than the TEST groups puppies. For the Yorkshire Terriers, there were no differences observed between the TEST and CONTROL groups.
  • Loud Noise Test: The results were identical to those of the novel item test. In addition, up to the age of 6 weeks, the Cavalier King Charles Spaniels showed more crouched posture behaviours than the German Shepherds or the Yorkshire Terriers. After 6 weeks, these differences became less apparent among all the puppies, where their behaviours in both the TEST and CONTROL groups were similar.
  • Cortisol: No salivary cortisol samples were taken from the Yorkshire Terriers so there is no data for this breed. The collection of samples from all of the Cavalier King Charles Spaniels and all the German Shepherds on every occasion was not always possible. Therefore, only the data for puppies from which 3 weeks or more of samples were successfully collected were used in the analysis. For both breeds, mean cortisol levels were higher at 4 to 5 weeks, then gradually declined through to 8 weeks of age. In addition, the German Shepherd puppies had higher mean cortisol levels than the Cavalier King Charles Spaniel puppies throughout the trial period. Compared to the CONTROL groups, the TEST group Cavalier King Charles Spaniel puppies had higher cortisol levels 20 minutes after each test than their pre-test values. This cortisol response did not occur in the German Shepherd puppies. The cortisol levels in the Cavalier King Charles Spaniels that showed fear-related avoidance behaviours were higher than in those that did not.

So, what can be learned from the results of this study? First and foremost, there are differences between different breeds of dogs in the age at which the onset of fear emerges –  earlier in German Shepherds and Yorkshire Terriers and later in Cavalier King Charles Spaniels. The results for the German Shepherds is consistent with those of a previous study in this breed, where at 5 weeks of age a significant fear response (onset of hazard avoidance behaviours) emerged in 90% of the puppies being studied, compared to Labrador Retriever puppies, where the figure was 4% at the same age (Coppinger and Coppinger, 2001). Incidentally, Ray and Lorna Coppinger are both COAPE Tutors, see the Tutors page on the COAPE website for more details.


For pet dogs in general, the average onset of hazard avoidance is 49 days, but it is just 19 days in wolf puppies (Coppinger and Coppinger, 2001). So, there is a link between the genetic process that has taken place to transform the wolf into the domesticated pet dog that not only affects appearance, but also behaviour. In particular, the timing of the onset of hazard avoidance behaviours in domesticated puppies has an enormous impact on the window of opportunity that is available – the critical period – to breed puppies that will become happy, healthy and emotionally well-balanced dogs.

Of all animals, dogs show an enormous variation in size within a single species, from less than 1 kg right up to nearly 100 kg, a one hundred fold increase. There is also much variation in body shape, with the most striking differences being in the shape of the skull. Skull shape in dogs varies from the dolichocephalic (long-nosed) breeds, such as Borzois and Whippets, at one end of the spectrum and the brachycephalic (short-nosed) breeds, such as Pugs and Cavalier King Charles Spaniels, at the other. Changes in the shape of the skull over evolutionary time inevitably leads to changes in the shape of the brain (Roberts et al., 2010), and this in turn changes behaviour (Coppinger and Coppinger, 2001).

As an aside, one of the consequences of shortening of the skull in the extreme through irresponsible breeding is the cause of Chiari-Like Malformation / Syringomyelia in Cavalier King Charles Spaniels and Bichon Frise.

In brachycephalic dogs, the eyes are located on the front of the face, as in humans, and this shifts the field of vision away from the sides and towards the front of the animal. This change is also associated with a reorganisation of cells in the retina leading to higher visual acuity in the frontal field (McGreevy et al., 2004). A study comparing the abilities of different types of dogs abilities to follow human pointing gestures, a measure of the development of canine social cognition and ability to cooperate with humans, has shown that brachycephalic dogs perform much better than their dolichocephalic counterparts (Gacsi et al., 2009).


COAPE - Centre of Applied Pet EthologyHead shape does not just affect a dog’s behaviour, it affects human behaviour towards dogs as well. Humans are innately attracted to infantile facial features such as bulging cheeks and large foreheads with low-lying and large eyes. These attributes induce nurturing feelings, especially in children and women, but also men (Archer and Monton, 2011; Borgi and Cirulli, 2013). Neoteny – the retention of juvenile physical and behavioural characteristics in the adult form – is an important attribute of canine domestication (Coppinger and Coppinger, 2001), and this feature is especially prominent in brachycephalic dogs. In many different species of domesticated animals including dogs, horses, cows, sheep, goats, pigs and rabbits, along with tameness, other physical attributes also emerge such as spotted/dappled coats and floppy ears. These physical changes come about as a result of a selective retardation of specific groups of cells during the development of the embryo (Trut et al., 2009). The breeders of the Cavalier King Charles Spaniels in this study commented that, compared to other breeds they were familiar with, in the Cavalier King Charles Spaniel puppies the eyes opened later and they also started the weaning process and exploration of their environment later as well. This observation was borne out by the fact that in this breed, testing was not begun until 5 weeks of age (4 weeks in the German Shepherds and the Yorkshire Terriers) because of the lack of response to stimuli.

In this study, as the 3 breeds of puppies grew up, it became apparent that the Cavalier King Charles Spaniels were left behind in terms of their general mobility on the test equipment. This occurred at 6 weeks old during the seesaw test and at 7 weeks old during the novel item and loud noise tests. The German Shepherd puppies were generally the most mobile of the 3 breeds from 6 weeks onwards. These observations fit in with the differences in behaviour found between dolichocephalic and brachycephalic breeds of dogs and the retention of neotenic characteristics in the latter (Coppinger and Coppinger, 2001; McGreevy et al., 2013).

All breeds of puppies in this study showed a pronounced crouching response to the noise test from 4 to 6 weeks old. This then declined after 6 weeks of age and few of the puppies showed this response at 7 weeks old. Scott and Fuller (1965) identified the emergence of a ‘startle response’ in their puppies at on average around 19.5 days old. They described this as a reflexive response that is not the same physiologically as a fear response. However, the influence of the effects of habituation as a result of repeated exposure to the loud noise cannot be ruled out here. Even if habituation was the case for some of the puppies, it did not affect the emergence of a measurable fear avoidance response (hazard avoidance) during the study period.

In the puppies where salivary cortisol was measured, the baseline levels where 2 to 4 times that found in adult dogs on average and were detectable from 4 weeks of age when testing started. Cortisol levels are used as an indicator of stress and anxiety in humans (Hellhammer et al., 2009) and in dogs (Glenk et al., 2013). Although higher than normal cortisol levels are associated with stress and anxiety in adults, this would appear not to be the case for the very young. In human babies for example, cortisol levels are very low in the first 14 days after birth. After 14 days old, exposure to any stressors causes a dramatic increase in the cortisol response. As the child grows up over the first 2 years, this response becomes less acute as fluctuations in cortisol normalise towards those of an adult (Drvaric et al., 2014). This is the normal developmental physiology of the adrenal glands and probably accounts for the large fluctuations in infantile cortisol found in this study. However, in this study, as the puppies grew older and a measurable fear avoidance response (hazard avoidance) emerged those Cavalier King Charles Spaniels that showed the response had significantly higher cortisol levels after the test than before the test compared to the Cavalier King Charles Spaniels that did not show the response.

This great study confirms the findings of other, earlier studies and gives us more useful information about the crucial milestones of puppy development.


© copyright Robert Falconer-Taylor, 2015

This article is an original work and is subject to copyright. You may create a link to this article on another website or in a document back to this web page. You may not copy this article in whole or in part onto another web page or document without permission of the author. Email enquiries to


PHOTO 2: Sprocker puppy at 18 weeks old, a rescue from Springer Rescue for Scotland, courtesy Robert FT.


Archer J. Monton S. 2011. Preferences for infant facial features in pet dogs and cats. Ethology, 117(3), 217-226.

Bekoff M. 2015. Judge Recognizes Two Chimpanzees as Legal Persons: A First. Psychology Today,  Apr 20, 2015 in Animal Emotion. Accessed 12 June, 2015.

Borgi M. Cirulli F. 2013. Children’s preferences for infantile features in dogs and cats. Hum. Anim. Interact. Bull, 1, 1-15.

Coppinger R, Coppinger L. 2001. Dogs: A startling new understanding of canine origin, behavior and evolution. Simon and Schuster.

Drvaric L, Waxman J, Van Lieshout RJ, Schmidt LA. 2014. Prematurity as a Context of Development at Risk. In Burack JA, Schmidt LA. (eds), Cultural and Contextual Perspectives on Developmental Risk and Well-Being, Chapter 9. Interdisciplinary Approaches to Knowledge and Development, Vol. 39. Cambridge University Press. ISBN 1-10700885-9.

Gacsi M, McGreevy P, Kara E, Miklosi A. 2009. Effects of selection for cooperation and attention in dogs. Behavioral and Brain Functions, 5:31.

Glenk LM, Kothgassner OD, Stetina BU, Palme R, Kepplinger B, Baran H. 2013. Therapy dogs’ salivary cortisol levels vary during animal-assisted interventions. Animal Welfare, 22(3), 369-378.

Hellhammer DH, Wüst S, Kudielka BM. 2009. Salivary cortisol as a biomarker in stress research. Psychoneuroendocrinology. 2009 Feb;34(2):163-71.

McGreevy PD, Grassi TD, Harman AM .2004. A strong correlation exists between the distribution of retinal ganglion cells and nose length in the dog. Brain, Behaviour and Evolution, 63 (1), 13–22.

McGreevy PD, Georgevsky D, Carrasco J, Valenzuela M, Duffy DL, Serpell JA. 2013. Dog behavior co-varies with height, bodyweight and skull shape. PLoS One. 2013 Dec 16;8(12).

McIntyre S. 2015. Animals are now legally recognised as ‘sentient’ beings in New Zealand. The Independent, May 17 May, 2015. Accessed 12 June, 2015.

Menteith C. 2014. The Puppy Plan. Kennel Club and Dogs Trust. Accessed 12 June, 2015.

Morrow M, Ottobre J, Ottobre A, Neville P, St-Pierre N, Dreschel N, Pate JL. 2015. Breed-dependent differences in the onset of fear-related avoidance behavior in puppies. Journal of Veterinary Behavior: Clinical Applications and Research. 2015 10(4) 286–294.

Roberts T, McGreevy P, Valenzuela M. 2010. Human-induced rotation and reorganization of the brain of domestic dogs. PLoS ONE, 5 (7).

Scott JP, Fuller JL. 1965. Genetics and the Social Behavior of the Dog. University of Chicago Press.

Trut L, Oskina I, Kharlamova A. 2009. Animal evolution during domestication: the domesticated fox as a model. Bioessays. 2009 Mar;31(3):349-60.