Development of Pecking in Ring Doves

Peter Balsam and James D. Deich
Department of Psychology, Barnard College of Columbia University 

I. Learning and Behavioral Development

Two relatively independent traditions in psychology emerged for the study of how experience affects behavior (Balsam & Silver, 1994). Developmentalists have long acknowledged a central role for the effects of experience on behavioral development and, quite independently, studies of learning and memory have occupied a central place in psychology. Yet there has been relatively little specific application of knowledge and principles in one domain to specific studies in the other domain. This lack of cross fertilization has occurred for a number of reasons (see Balsam & Silver, 1994) but, most importantly, the developmentalist has been most deeply concerned with the emergence of new response forms over time (Kagan, 1971; Lewis & Starr, 1979; Werner, 1957), whereas the learning psychologist has been most concerned with changes in controlling stimuli rather than changes in response topography (Balsam, 1988). However, it is clear that both areas have identified and analyzed important processes and that a complete account of both behavioral development and learning will require a synthesis of these areas (Balsam & Silver, 1994).

 In this chapter we focus on how learning principles can be used to understand the development of behavior by describing our work on the development of pecking in the ring-necked dove (Steptopelia risoria). We study this system as a model for understanding how learning contributes to behavioral development (Balsam & Silver, 1994).  First, we describe the character of adult pecking.  Next, we describe how this response emerges during development followed by a summary of our research on how birds learn what to peck at and how they learn to make the skilled adult motor response.  Lastly, we return to the general implications of this work for an analysis of behavioral development.

II. Pecking

The beak is the primary means by which most birds influence the world. It is used for prehension in a similar way to how primates use fingers and hands to reach, grasp, and manipulate objects. The beak is used to pick up and ingest food, preen the feathers, and build the nest. In many altricial species the beak is used to feed the young, either by grasping the beaks of the young and delivering food regurgitatively or by using the beak to seize food and carry it back to the nest. Additionally, the beak is sometimes used to drink, inflict injury, and to produce song. All of these actions involve different topographies of movement that are tuned to the specific stimuli and functions. These behaviors are highly skilled in adults.

For example, the video clip illustrates how picking up a piece of seed is a highly  skilled action. The views from above and below expose the precision of the movement.   An accurate response requires that information about the spatial location and properties of the target seed be rapidly acquired. This information is then used to precisely guide the motion of the head and the coordinated movement of the beak in grasping a seed. This is a remarkably flexible system in the fine adjustments that are made during each unique occurrence of the behavior. The trajectory of the head must change with each change in body position and the opening of the beak must be scaled to each change in seed size.  In adults, the behavior is stereotyped and efficient, but it does not start out that way.

Prehension in avians, like that in primates (Jeannerod, 1981, 1984), requires the coordination of two response components (Deich, Allan, & Zeigler, 1988; Klein, Deich, & Zeigler, 1985;  Zeigler, Levitt, & Levine, 1980;): the effector must be transported to the target object and then closed around it. In primates the transport component is typically called reaching or simply transport; for avians we prefer the term thrust. Closing around and securing of the target is called grasping. A successful peck requires that thrusting and grasping be well coordinated.

The duration of thrusting and the accompanying opening phase of grasping is so short (about 50-120 ms, Deich, Klein, & Zeigler, 1985; Bermejo, Houben, Allan, Deich, & Zeigler, 1989) in the Columbidae that either high speed motion picture photography (Klein, Deich, & Zeigler, 1985) or direct transduction by attached sensors is necessary to measure these movements  precisely . A system for the direct measurement of the gape component was developed in the pigeon (Deich, Houben, Allan, & Zeigler, 1985). A small rare-earth magnet is glued to the bird's lower beak and a small magnetic sensor to the upper beak. The intensity of the magnetic field detected by the sensor decreases as the beak opens. The signal from the sensor is input into a computer and the changes in interbeak distance are recorded in real time.

We employed this system to record the gape of adult ring doves (from Deich & Balsam, 1994). This figure shows the movement of the head synchronized with that of the beak. The gape sensor record, shown in the lower panels, is also synchronized with the drawings. The top panel of the gape sensor record shows displacement and the bottom panel velocity. Negative velocities indicate reclosure of the beak. As an ingestive sequence begins, the head is briefly fixated above the seed (Goodale, 1983) and then drives downward toward the seed. During this thrusting movement the beak opens.   Gape reaches its maximum (1-2mm wider than the seed) while the head is being thrust downward. As the beak reaches the seed it is reclosed around it creating a plateau in the record. After reclosure with the seed secured at the beak tip, there is a set of higher velocity beak opening movements that typically have only a single velocity peak and result from moving the seed to the back of the beak for swallowing. The movement of the seed in the beak is called mandibulation, and is often assisted by the tongue which sticks to the seed and pulls it backwards ( Van Gennip, 1988; Zweers, 1982b). At the end of the ingestive sequence the beak opens one final time. This opening appears on the gape record as a peak with rounded shoulders. According to Van Gennip (1988), this epoch of the signal is caused by movement of the tongue to allow swallowing. 

III. Development of Pecking

The Columbidae family (pigeons and doves) provides good model of how adult feeding responses emerge during development (Balsam, Deich & Hirose, 1992). The general developmental pattern is similar in all members of this family and has been described in detail for the ring dove (Lehrman, 1955; Silver, 1978; Wortis, 1969). 

After a male and female ring dove have mated, the female typically lays a clutch of two eggs that are incubated by both parents for 14-15 days prior to hatching. Initially, the young are unable to feed themselves. The parents feed the squab "crop-milk" (Wortis, 1969), a cheese-like substance produced in the crop. Both parents participate in the regurgitative feeding of the young (see a video of this behavior). These feedings are initiated by the parents after hatching, but by the time the young leave the nest on about post-natal day 10 (PD 10), the squab beg vigorously to obtain parental feeding. As can be seen in the next video clip, the begging consists of the squab thrusting its beak at the parent's beak while making very rapid fluttering motions with its wings. The begging is sometimes accompanied by a chirping-like call.  During regurgitative feeding, the parent grasps the bill of the squab between its mandibles and makes vigorous pumping movements with its upper body, particularly with its neck and head. From around the third day post-hatching until the squab begin getting food on their own, the parents feed the squab crop-milk mixed with increasing amounts of seed (Graf, Balsam & Silver, 1985). The amount of crop-milk fed to the squab increases through about PD 5 and declines thereafter. The amount of seed fed to the squab steadily increases until they are eating seed on their own during the fourth week after hatching.

The squab's and parents' behavior surrounding feeding interactions changes during the transition  from dependent to independent feeding (Hirose & Balsam, 1995). All of the young that we have observed will beg for food at least through the time of weaning (PD 21).   The next figure shows that  Parental feeding (see triangles) of young begins to decline near the end of the third week and usually ceases by the end  of the  fourth week post-hatch. Pecking begins around PD 14 and increases in frequency. Squab begin to successfully ingest seed by about PD 16 and continue to improve in efficiency (see circles). By PD 28 they are nearly as efficient as adults at ingesting seeds (Balsam et al., 1992). We (Deich, Tankoos & Balsam, 1995) used the hall-effect system to follow the changes in gaping from the time squab start pecking until they independently feed. There were three types of gapes that were dominant during different steps in the development of feeding. Over 90% of all pecks could be sorted into the three types shown in the next figure. All of the squab moved through the same sequence of changes in topography as pecking became more efficient. Each transition in form was  associated with improved efficiency of pecking. These were just the sort of changes that one might expect if the squab’s experience with seed was contributing to the changes in response form.

Thrusting and gaping

In our first approach to the question of whether normal development of pecking depends on experience, we (Graf et al., 1985) reared squab without any exposure to seed. We achieved this by grinding seed into a fine powder and gradually making this the only source of food available to parents prior to hatching. This powdered seed was also the only source of food available to the families after the squab hatched. We have found no differences between these powder-reared squabs and seed-reared squabs in growth and general behavior. Beginning on PD 14 and on all subsequent days, squabs were put into a chamber with seed on the floor for twenty minutes. We found that powder-reared squabs pecked very little during these test sessions as compared to seed-reared subjects indicating that direct experience with seed is important for development of the adult response. 

We then began to analyze whether and what kind of experiences are necessary for normal development to occur.  First, we have found that 2-3 week old squabs have an unconditioned tendency to thrust at seed (Graf et al., 1985). This response will habituate unless the sight of seed is followed by positive ingestional consequences (Balsam et al., 1992). These Pavlovian pairings increase the rate of thrusting at seed (Deich & Balsam, 1993) and these pairings occur in natural parent-squab interactions (Hirose & Balsam, 1995; see Akins & Hamilton (2001) for the role of Pavlovian conditioning in avian sexual behavior). 

The pairings of seed and food, while increasing head thrusts toward seed, do not result in the release of the skilled adult response. Initially, the opening and closing of the beak is only loosely coupled to head movement (Deich & Balsam, 1993; Deich, Tankoos & Balsam, 1995) and seems to be elicited by the head movement (Wall, Saluja, Iskrant, Pierson, Deich & Balsam, 1996).  These gapes are not scaled to seed size and there is limited accuracy in the targeting of the peck.  Suspecting that feedback from successful and unsuccessful pecks was necessary for sustaining pecking and for shaping the squab's behavior toward the adult form, we turned our attention to analyzing the roles of feedback in the development and maintenance of pecking.

Role of feedback

We guessed that feedback from the beak during the movement and feedback about the consequent success and failure of pecks was essential for the skill to develop. We thought that the feedback might be important for sustaining and strengthening the peck response.   Additonally, unless the mapping from the visual cue to the size of the gape was "hard-wired", the afferent feedback from particular gapes must be associated with the success and failure of pecks at specific targets. In two different lines of experiments we have examined the role of feedback on the development of pecking. 

To examine the role that feedback plays in sustaining the overall level of pecking, we exposed a group of powder-reared squab to seed for 20 minutes each day during the second week post hatch, but did not allow them to experience the seed in their beak. We did this by gluing the seed to the floor of the test chamber. These squab were immediately fed by the experimenter after the test session. The level of pecking observed in this group on day 22 when they were tested with loose seed on the floor was substantially less than the pecking observed in a second group who was given unglued seed throughout the prior training and testing days. Subjects exposed to the glued seed and deprived of feedback provided by successful pecks responded considerably less than the subjects that got to handle seed with the beak during testing. In fact, the glued-seed subjects pecked at exactly the same level as subjects that only received pairings of the sight of seed with food (Deich & Balsam , 1993; 1994).   Pavlovian pairings promote pecking but the opportunity to handle seed results in a higher peck rate.  Thus one role of the feedback is to increase the overall rate of pecking. It is plausible that when parents regurgitatively feed offspring the sensation of seed in the beak becomes a conditoned reinforcer because of its association with positive nutritional consequences.  Once the squab starts pecking  successful responses provide this immediate reinforcement which would lead to a higher peck rate. 

We began to study the effects of feedback in more detail in a collaboration with H.P. Zeigler's group at Hunter College. Zeigler and his colleagues have studied the role of trigeminal input on feeding in pigeons (Zeigler, Bermejo & Bout, 1994). By sectioning some or all of the trigeminal branches, they have found that trigeminal input plays a role in motivation, spatial accuracy, and the efficiency of grasping and mandibulation. 

The afferent feedback from the beak is carried by the three branches of the trigeminal nerve: maxillary, mandibular and ophthalmic branches. We (Ye, Wild, Balsam & Zeigler, 1998) examined the circuit that carries the feedback in adults and squab that were 24-48 hours old. We found that the organization of the trigeminal ganglion, its somatotopic projections upon the principal sensory nucleus (PrV) and the projection to the telencephalon are similar in the adult and hatchling. Consequently, developmental changes in pecking are unlikely to be due to maturation of circuitry but rather are driven by the information carried in the system. 

We have begun to look at the role of the trigeminal input in more detail by reducing the input from the beak. We first did this in adults by sectioning the peripheral nerve. The most fascinating of these results is revealed in the type of errors made after deafferentation. The link shows the types of errors made by an adult prior to and after bilateral ophthalmic sectioning. The errors increase after the surgery, but notice that the immediate effect is to produce a small increase in grasping errors. In these cases, the bird made contact with the seed but could not hold it in its beak. Subsequently grasp errors decline but spatial targeting errors increase. The bird misses the seed toward which its peck is directed. This demonstrates that the trigeminal input plays the important role of giving feedback about the success versus failure of the spatial aspects of the peck. When that feedback is attenuated the targeting becomes more variable. Sham controls do not show decreased efficiency of eating, indicating a specific effect of deafferentation. 

We have also sectioned the trigeminal a few days after hatching and allowed the squab to be reared by their parents. They grow normally and interestingly, they all eventually learn to eat seed. However, the rate at which they learn is very much affected by the quality of the sensory feedback from the beak. The eating efficiency of squab that have intact trigeminal nerves improves much faster than the efficiency of squab who have one or two branches of the nerve severed on day 3 post-hatch. The development of skilled pecking is highly dependent on this iput. The more attenuated the input the slower the learning. We do not know if squab could eventually learn to handle food without sensory input from the beak but this input clearly plays a key role in normal development. There are two likely components to the deficit produced by deafferentation. The reduced feedback from the beak movement may interfere with learning to scale the gape to seed size. Additionally, reduced feedback from seed in the beak is likely to make it more difficult to distinguish successful and unsuccessful pecks. Both uses of the feedback seem essential for the skill acquisition. 

In sum, feedback about success and failure seems to contribute to every aspect of the skill.  It contributes to the overall level of pecking and affects the targeting, grasping, and handling of the seed.  When normal feedback is present during development the highly flexible adult response emerges.  The bird pecks very precisely at a wide range of seeds from a range of positions.  We wondered what learning underlies this flexibility.  How does the bird build such a flexible repertoire?

Scaling of Gape to Seed Size

We already knew that maturation alone could not account for the development of the adult gaping response (Deich & Balsam 1994). Squabs reared on the powder diet do not show the adult gape form even when tested at an age when seed-reared squabs are quite proficient. The skilled adjustment to seeds of different size exhibited by adults (Deich & Balsam, 1994) depends on experience. Thus we turned our attention to identifying the specific experiences necessary for this aspect of the adult skill. 

Our doves are usually raised on a standard dove mix. This mixture contains seeds of many sizes and shapes which range from about 1mm (millet) to about 12 mm (peas). It is possible that doves must separately learn to handle all of the different seeds before they show the precise and flexible adult response. Alternatively, the doves may acquire a generalized motor program. One characteristic of a generalized motor program is that quantitative adjustments of behavior occur in novel situations. For example, once we are skilled walkers we adapt to new terrain with little loss of skill.  Similarly, we adjust the opening phase of our grasp to the width of any new object we pick-up. Presumably, once the motor program is functioning we are able to extract the necessary information to use the program from visual cues provided by the novel object. 

It was an open question as to whether experience is necessary for the development of motor programs. Though skill surely improves with experience, but it is not clear whether or not the initial induction of motor programs requires experience. Because we can limit the doves’ experience with seed during development we can study this question with respect to pecking. We know from our powder-reared subjects that they are not able to pick up seed when they first encounter it – even at an age when normally reared subjects would be quite proficient (Deich & Balsam, 1994). Perhaps experience with a single seed would be sufficient for inducing the motor program. Alternatively, experience with two or more points along a dimension may be necessary for the development of a generalizable skill. 

To test this hypothesis, we compared the pecking of normally reared doves to a group reared on a single size pellet.   In the group reared on a single size pellet their diet consisted solely of pellets, 2 mm in diameter. This was the only food available to the parents and squab until the time of weaning and the only food available to the squab until they were 18-24 months old. A second group of subjects was reared on mixed grain until they were between 18 and 20 months old. This group was switched to a diet of only 2 mm pellets by gradually increasing the proportion of their food that consisted of pellets over the course of a week. These subjects were then maintained solely on these pellets for at least two months prior to testing. 

On the first test day, subjects were given 1 mm, 2 mm, and 3.6 mm pellets to eat. On test trials ten seeds of each size were presented. All pecks were videotaped and scored for accuracy.  The set-up was similar to the one illustrated in the video.  The seeds and beak were visible from below and from the side of the test aquarium.  We classified the types of errors made by the birds into the six categories described in the figure showing the data from this test.  The result of this test showed that subjects reared on mixed seed (not shown in the figure) had little trouble handling the 1 and 3.6 mm pellets even though they had only handled 2 mm seeds in the recent past. The subjects who had only experienced a 2 mm diet (shown in the figure) exhibited a different pattern of results during the generalization test. These subjects had little trouble with the 1mm seed. They made a few errors in which they contacted the seed but failed to grasp it.  This was about the same number and type of error that they made when handling the very familiar 2 mm pellet. In contrast, the subjects reared on the restricted diet did very poorly on the 3.6 mm seed. They often contacted the seed and failed to grasp it or got their beaks around it briefly until the pellet squirted out as the beak closed.  They were not scaling the gape to the size of the seed or positioning the beak to securely grasp the seed.  It seems likely that the gape they had learned for 2mm seeds was wide enough and targeted well enough to handle the smaller, 1 mm, seed but not adequate for the larger seed. Over the next few days they readily learned to handle the larger seeds. In fact, they made very few errors after 50 or so experiences with the 3.6 mm seed. After a week of this testing, we introduced a 7.2 mm pellet during testing. None of the birds had any trouble with the new seed. Apparently, it takes experience with two or three points on a dimension to induce the generalizable skill.

Synopsis: Learning and the development of pecking

There is an unconditioned tendency for squabs to thrust at grain-like objects (Balsam et al., 1992). However, this response habituates quickly unless the squab experiences Pavlovian pairings of the sight of grain with feeding (Balsam et al, 1992; Deich & Balsam, 1994). The Pavlovian induced thrusts do not have the coordinated gape found in the adult peck (Deich & Balsam, 1994; Deich, Tankoos, & Balsam, 1995). There is only a loose coordination of gape and thrust, induced by the thrusting motion, itself (Wall et al., 1996). Squabs must have experience with handling and ingesting seed for the adult topography and coordination to emerge (Deich & Balsam, 1994). Through differential feedback from successful and unsuccessful pecks, the gape component and its coordination with the thrust component are moved toward the more effective adult form by a process of response shaping. Operant conditioning is proposed as the underlying operative process because neither maturation nor extensive Pavlovian training is sufficient to produce the stereotyped adult response (Deich & Balsam, 1994). Furthermore, reduction in the oro-sensory feedback as a result of beak deafferentation interferes with the development of the coordination. 

All of these processes are important in the normal developmental context. From the time that parents start to feed squab seed the feel of seed in the mouth becomes associated with positive nutritional consequences. During the second week after hatching the parents stop consistently feeding the begging squab and, instead, go to peck at seed themselves (Wortis, 1969). The squabs follow the parents to the seed and make some unconditioned thrusts at food. Additionally, we have found that squabs tend to peck when their parents do (Iskrant & Balsam, 1994; see Zentall & Akins (2001) for more on social learning) and have documented that this social coordination is present in the undisturbed interactions of parents and squabs (Hirose & Balsam, 1995). The unconditioned tendency and the social enhancement guarantee the squabs visual exposure to seed. During this stage, parents generally feed squabs within a minute or two of this exposure (Hirose & Balsam, 1995). This allows for the sight of seed to be paired with food. The frequency of thrusting toward seed is increased by the Pavlovian pairings (Balsam et al., 1992). The squab’s thrusting movement elicits loosely coordinated gapes (Wall et al., 1996). Some of these thrust-gape coordinations result in seed in the beak. Feedback from the seed in the beak reinforces the successful actions. The gape component and its coordination with the thrust component are moved toward the more effective adult form by a process of response shaping by reinforcement. 

IV. Learning and Behavioral Development 

The highly skilled and adaptable adult pecking response is produced by an intricate interplay between the squab’s reflexes, the teaching of the parent, and the very precise feedback that is provided by the environment for success and failure.  In these regards the development of pecking shares many features of behavioral development that are common to altricial species. 

Within the context of the parent-offspring interaction there is typically a great deal of flexibility in development as a result of experience.  For example, in the case of feeding, local conditions of food availability can determine the rate at which weaning occurs, the specific foods that a juvenile learns to find and to consume, and the specific responses that are used for foraging and eating (Balsam et al., 1992).  Thus experiences are a critical influence on development.  An important question is how to understand the specific ways in which experience affects these behaviors.  

A number of developmentalists have articulated the view that laboratory learning principles do not capture the ways that experience affects development.  Gottlieb (1976, p. 232) wrote, “Traditional forms of learning (habituation, conditioning and the like) have not proven very useful in explaining the species-typical development of behavior…” In this context, we view the studies described in this paper as a true test of validity for laboratory learning principles and believe that they have done pretty well. At least in the case of pecking, laboratory-derived learning principles have served us well by providing an analytic tool for understanding how specific experiences contribute to the development of specific aspects of behavior.

Perhaps, the major contribution of learning principles is to conceptualize development in terms of separate stimulus and response transitions.  Learning theory can provide the analytic tools for understanding transitions in stimulus control  (Alberts, 1978; Alberts & Brunges, 1979; Gottlieb, 1983; Hall & Swithers-Mulvey, 1992) by looking for things like changes in predictive validity, blocking, overshadowing and potentiation during development.  Likewise, the use of learning principles to understand response transitions is also fruitful (Balsam, Deich, Ohyama & Stokes, 1998). 

The emergence of new behavior is also a significant aspect of development that is influenced by experience.   New response forms can be induced by elicitation, Pavlovian conditioning, habituation, and shaping (Balsam & Silver, 1994).  Again the learning principles seem useful for understanding the ways in which experience accomplishes these changes in development.  Furthermore, one can conceive of the transitions in functionally equivalent behaviors as the product of changing relative reinforcement rates.  For example, the transition from dependent to independent feeding is likely modulated by the pay-off that each affords. The selection of specific feeding topographies is likely determined by the successes and failures of those actions in the past. In sum, we believe that learning principles provide a very useful framework for understanding the roles of experience in the ontogeny of behavior.

V. References

Alberts, J. R. (1978). Huddling by rat pups:  Multisensory control of contact behavior. Journal of Comparative and Physiological Psychology, 92, 220-230.

Alberts, J. R. & Brunjes, P. C. (1979).  Ontogeny of olfactory and thermal determinants of huddling in the rat.  Journal of Comparative and Physiological Psychology, 92, 897-906. 

Balsam, P. D. (1988) Selection, representation and equialence of controlling stimuli.  In R. Atkinson, R. J. Herrnstein, G. Lindzey, & R. D. Luce (Eds.) Handbook of experimental psychology. New York, NY: Wiley. 

Balsam, P., Deich, J. & Hirose, R. (1992).  The roles of experience in the transition from dependent to independent feeding in ring doves.  In G. Turkewitz (Ed.), Developmental psychobiology (pp 16-36).  New York, NY:  New York Academy of Sciences.

Balsam, P. , Deich, J., Ohyama, T. & Stokes, P.  (1998).  Origins of new behavior.  In W. O’Donohue (Ed.)   Learning and behavior therapy ( pp 403-420). Boston, MA: Allyn & Bacon.   

Balsam, P., Graf, J. S. & Silver, R. (1992).  Operant and Pavlovian contributions to the ontogeny of pecking in ring doves. Developmental Psychobiology, 25, 389-410.

Balsam, P. D. & Silver (1994).  Behavioral change as a result of experience:  Toward principles of learning and development.  In J. Hogan & J. Bolhuis (Eds.), Causal mechanisms of behavioral development. Cambridge, MA: Cambridge University Press. 

Bermejo, R., Houben, D., Allen, R. W., Deich, J. D., & Zeigler, H. P. (1989). Prehension in the pigeon I: descriptive analysis. Experimental Brain Research ,  75, 569-576. 

Deich, J.D., Allan, R.W. & Zeigler, H.P. (1988).  Conjunctive differentiation of gape during food-reinforced keypecking in the pigeon. Animal Learning & Behavior, 16, 268-276.  

Deich, J. & Balsam, P.  (1993).  The form of early pecking in the ring dove squab. (Streptopelia risoria).  Journal of Comparative Psychology, 107, 261-275.

Deich, J. & Balsam, P. (1994).  Development of prehensile feeding in ring doves (Streptopelia risoria).  In M. Davies & P. Green (Eds.), Perception and motor control in birds (pp. 160-181).  Heidelberg, Germany: Springer-Verlag .

Deich, J. D., Houben, D, Allan, R. W., & Zeigler, H. P. (1985) "On-line" monitoring of jaw movements in the pigeon.  Physiology & Behavior, 35, 307-311.  

Deich, J. D., Klein, B. G., & Zeigler, H. P. (1985). Grasping in the pigeon: Mechanisms of motor control. Brain Research ,  337 , 362-367. 

Deich, J., Tankoos, J. & Balsam, P.D (1995).  Ontogenetic changes in the form of the gaping at seed in the ring dove (Streptopelia risoria). Developmental Psychobiology, 28, 147-163.

Goodale, J. (1983).  Visually guided pecking in the pigeon (Columba livia).  Brain Behavior Evolution ,  22 , 22-41.

Gottlieb, G. (1976). Conceptions of prenatal development: Behavioral embryology.  Psychological Review, 83, 215-234.  

Gottlieb, G. (1983).  The psychobiological approach to developmental issues. In Handbook of child psychology (Vol. 2).  P.H. Mussen (Ed.).  New York, NY: John Wiley & Sons. 

Graf, J.G., Balsam, P.D., & Silver, R.  (1985). Associative factors and the development of pecking in the Ring Dove. Developmental Psychobiology, 18, 447-460.  

Hall, W. G. & Swithers-Mulvey, S. E. (1992).  Developmental strategies in the analysis of ingestive behavior. In G. Turkewitz & (Ed.). Developmental psychobiology. Annals of the New York Academy of Sciences, 662, 1-15. New York, NY: New York Academy of Sciences.

Hirose, R., & Balsam, P. (1995). Parent-Squab interaction during the transition from dependent to independent feeding in the ring dove  (Streptopelia risoria). Animal Behaviour, 50, 595-606. 

Iskrant, K. & Balsam, P. (1994). Social and Nutritional influences on feeding in ring dove squab.  Developmental Psychobiology, 27, 195-204.

Jeannerod, M. (1984).  The timing of natural prehension movements.  Journal of Motor Behavior, 16, 234-254.  

Kagan, J. (1971). Change and Continuity in Infancy. New York, NY: Wiley. 

Klein, B.G., Deich, J.D. & Zeigler, H.P. (1985). Grasping in the pigeon: final common path mechanisms.   Behavioral Brain Research ,  18 , 201-213.  

Lehrman, D.S. (1955). The physiological basis of parental feeding behavior in the ring dove (Streptopelia risoria). Behaviour , 7,  241-186.  

Lewis, M. & Starr, M.D.  (1979).  Developmental continuity. In Handbook of infant development, J. Osofsky (Ed.) (pp.653-670). New York, NY: Wiley 

Silver, R. (1978). The parental behavior of ring doves. American Scientist, 66, 209-215

Van Gennip (1988). A functional-morphological study of the feeding system in the pigeon (Columba livia L.): behavioral flexibility and morphological plasticity.  Unpublished doctoral thesis, University of Leiden.

Wall, E., Saluja, M.,  Iskrant, K.,  Pierson, E.,  Deich, J.  & Balsam, P. (1996, March). Origins of gaping in Ring Doves. Annual Conference of the Eastern Psychological Association, Philadelphia, PA.

Werner, H. (1957).  The concept of development from a comparative and organismic point of view.  In D.B. Harris (Ed.), The concept of development.  Minneapolis, MN: University of Minnesota Press.

Wortis, R.P. (1969).  The transition from dependent to independent feeding in the young ring dove. Animal Behavior Monographs, 2, 1-54. 

Ye, S., Wild, J. M., Balsam, P. & Zeigler, H. P. (1998). Organization of quinto-frontal sturctures in the hatchling ring dove (Streptopelia risoria). Brain Research, 788, 349-352.

Zeigler, H. P., Bermejo, R. & Bout, R.  (1994). Ingestion, prehension and the sensorimotor control of the jaw.  In M. Davies & P. Green (Eds.), Perception and motor control in birds.  Heidelberg, Germany: Springer-Verlag.

Zeigler, H. P., Levitt, P., & Levine, R. R. (1980). Eating in the pigeon (Columba livia): Movement patterns, stereotypy and stimulus control. Journal of Comparative Physiological Psychology, 94, 783-794. 

Zweers, G. A. (1982). Pecking of the pigeon (Columbia livia L.). Behaviour, 81, 173-230.

Acknowledgments  We thank all of our many students that contributed to the work on the project. We also add a special thank you to Alice Deich who helped in many ways. The work was supported by NSF grant IBN-9727428.