Reconstructing the behaviors of extinct species: An excursion into comparative paleoneurology


  • Scott W. Rogers

    Corresponding author
    1. Salt Lake City VA-Geriatrics Research, Education and Clinical Center and Department of Neurobiology and Anatomy, University of Utah School of Medicine, Salt Lake City, Utah
    • Neurobiology and Anatomy; MREB 403, University of Utah School of Medicine, 20 North 1900 East, Salt Lake City, UT 84132-3401.
    Search for more papers by this author


How can the behavior of an extinct species be reconstructed—say a dinosaur such as Allosaurus? Despite the relatively abundant fossilized remains of this and other dinosaurs, the incompleteness of the fossil record has permitted room for considerable speculation, mythology, and perhaps a bit of unsettling reflection on what factors contributed to the eventual fate of these remarkably successful animals. Among the speculations is how these ‘bigger-than-life’ creatures behaved, a topic that itself can attain equal diversity and grandeur. With recent advancements in measuring the relatedness of living organisms, how genetics contribute to brain development and how this relates to behavior, combined with the availability of newly discovered high quality fossils and imaging methods to exploit their secrets, novel insights into how extinct creatures such as Allosaurus intermingled with its many relatives over 100 million years ago are beginning to emerge. © 2005 Wiley-Liss, Inc.

The fossil record, not unlike an aging photo album, offers a fragmentary glimpse into the existence and life styles of the many great and small creatures that once roamed this planet as a part of its common life history. Prominently placed within those records are the dinosaurs who lived for some 100 million years, a testament to their remarkably effective approach to living in their surroundings. However, beyond their bones, we are left with little hope of acquiring evidence of their genetic constitution since DNA does not fossilize nor is it preserved [Nielsen-Marsh, 2002]. Attempts to reconstruct their behavior seem equally discouraging because this understanding ultimately requires examination of the brain, which of course does not fossilize. Or does it? In recent years there has been the recognition that soft tissue structures can indeed fossilize and that the appraisal of these specimens using new methods available through the rapid developments of computer assisted medical imaging or the approaches to chemical analyses of great sensitivity such as mass spectroscopy. Further, combining these technological advancements with considerable advances in our understanding of comparative genomics and neuroanatomy as will be discussed here is offering new insights into the fossil record, including the brain and the likely behaviors of extinct creatures that contributed significantly to their long-term success as well as their ultimate failure.

An amazing fact is that soft tissues can leave their mark in fossils and in some cases they have been overlooked almost certainly because of the notion that it is unnecessary to look for what you don't expect to find. In fact, over 100 years ago early paleontologists recorded the first evidence of fossilized ‘soft tissue’ structures when a well-preserved feather was found in sediments in Germany that had fallen from a Jurassic Archaeopteryx some 80 million years earlier. In the early 20th century Walcott and coworkers [see Gould, 1990] described fossils harboring detailed and exotic soft-tissue structures that were preserved when a sampling of the abundant Cambrian life forms were buried by fine sediment at the bottom of a very deep, cold, and deoxygenated oceanic trough that some 500 million years later would be named the Burgess shale. More recently, many exquisite fossils with fine soft-tissue details were described that range from preservation of proteins [Gurley et al., 1991] to internal organs including the digestive and cardiovascular systems of dinosaurs [e.g., Sasso and Signore, 1998; Ruben et al., 2003].

But the brain is a special case. This organ is subject to rapid degradation after death, and scavengers rarely leave it intact to assist the already rare processes leading to fossilization. It is in this context that we turn our attention to Allosaurus fragilis (Fig. 1a), the dominant carnivore of the upper Jurassic period. It is a conspicuous member of the massive radiation of Archosuars that gave rise to Crocodilia (including crocodiles, caimans and alligators—the latter term being used here to collectively describe this group), theropod dinosaurs (including Allosaurus and Tyrannosaurus) and birds. We are afforded a remarkably abundant glimpse of this extraordinary creature due to the discovery in a small area in central Utah termed the Cleveland-Lloyd quarry where the fossilized remains from some 40 Allosaurids that were rapidly buried in a fine silty mud that became their collective sarcophagus were found [Madsen, 1976; Stokes, 1985]. The simultaneous presence of one or two plant-eating dinosaurs suggested that this extraordinary gathering of carnivores also recorded a fossilized behavior; that of a predator trap where a wayward herbivore became entrapped in mud leading to cries, struggle and the eventual stench that attracted Allosaurids who were in search of easy prey or carrion who themselves also became trapped and entombed [Madsen, 1976]. The infiltration by the fine barium-rich mud and the fossilization that ensued also preserved an exceptional Allosaurid endocranial cast (Fig. 2a–c) that recorded the sensory organs, blood vessels, cranial nerves, and sufficient internal detail to be examined further using spiral CT-scanning imaging [Fishman and Jeffrey, 1995; Rogers, 1998]. This finding, placed in terms of modern genetic and phenotypic analyses, suggests that we may be able to reconstruct the behaviors of these creatures with increased precision to clearly resolve how they dominated their surroundings for some 40 million years.

Figure 1.

Panel A shows a silhouette of Allosaurus with its skeleton and a six-foot tall hominoid standing nearby. The positioning of the head as level to the ground is based on the location of the endocast semicircular canals (text and Rogers [1998]). In this Figure, an adult Allosaurus was approximately 20 feet tall and weighed about 2 tons. Panel B. The currently accepted cladistic relationship between key evolutionary groups discussed in the text. The cladogram is adapted from Wilkins [2002] and shows the relatively close relationship between members of the Archosaurian lineage including alligators (Allig), dinosaurs (Dinos), and birds (Aves). Mammals (Mam), indicated by an erect hominoid, branched from the line that gave rise to Archosaurs several nodes earlier than was previously suggested [Gardiner, 1982] from phenotype-based cladograms where mammals and birds shared a close lineage as did alligators and reptiles (Rept). In both cases amphibians (Amph) represent a common but very distant ancestor. Panel C. Alignment of brains from key species showing the progressive enlargement of the telencephalon (double arrows) relative to hindbrain structures such as the cerebellum (asterisks) whose location has remained relatively constant. The change in size of the primary olfactory processes (single arrow) relative to the telencephalon is noted. In humans the olfactory processes are greatly reduced in size and are located on the posterior aspect of the telencephalon not visible from this aspect.

Figure 2.

Panel A shows the right side of the Allosaurus endocast recovered from the Cleveland-Lloyd quarry by James Madsen [Madsen, 1976]. It is 16 cm tall, 16 cm long and narrow, being only 4 cm wide at its broadest point [Rogers, 1998, 1999]. A conspicuous olfactory system is well defined where prominent olfactory bulbs (OB) are separated from the endocast cerebral hemisphere (CH) by the intervening constriction of the olfactory peduncles (op). The bulge at its greatest lateral extension defines the cerebral region of the endocast, before it again tapers sharply in the caudal direction near the cerebellar (cb) region and the medulla/brain stem (BS). At the cerebellar base the ophthalmic and maxillary-mandibular branches of the trigeminal [tg] nerves as well as the optic nerve (on) and oculomotor and/or trochlear nerves (behind the optic nerve) are present. Blood vessels are also present including the prominent lateral sinus (ls). Also noted is flocculus (fl), and extension of the cerebellum that is surrounded by the semicircular canals (SC) of the vestibular apparatus. The specimen is mounted on a metal post (p). Panel B shows the image generated through computer reconstruction of the spiral computed tomography image collected with a GE Light Speed Plus CT scanner [see also Rogers, 1998]. The image is a computer rendered reconstruction of images taken of the endocast in approximately the same plane as shown in Panel A. In this image the surface material has been made transparent and colored light orange to reveal the internal endocast structure. Gray colored material is believed to reflect infiltrate that is consistent with blood vessels, dura and material in which the brain density (colored) is located. This image, which is of much greater resolution and detail than previously reported [Rogers, 1998, 1999] shows the majority of the telencephalon (CH) and the brain base in green while the cerebellar region (Cb) is in blue. Note the region caudal to the telencephalon (colored darker green) is consistent with the diencephalon region that in earlier scans was not visible and thought to be lost [Rogers, 1998]. Notably the density lies well centered in the endocast immediately caudal to the olfactory processes consistent with the location of the telencephalon and it does not fill the endocast consistent with modern archosaur brain-endocast relationships such as the alligator (see text). The gray material is composed of blood vessels, dura and structure resembling connective tissue that surrounds the base of the brain (see also Rogers, [1998, 1999]). This is interrupted just caudal to the most visible rostral aspect of the cerebral hemisphere by a drill site used to mount the post in the original mounting of the specimen. Panel C. The endocast is oriented at the most rostral aspect (OB) looking toward the posterior. To the right is the CT image from this aspect colored as in Panel B. From this aspect the cerebral hemisphere of the telencephalon (CH, green) is clearly visible elsewhere it is symmetrically oriented within the surrounding supporting material. Again, the surface of the endocast has been rendered transparent and is in dark orange. Panel D. The endocast is placed over a skull of Allosaurus to show its relative orientation. Note the orientation of the horizontal canal of the vestibular apparatus (see below) strongly suggest that Allosaurus walked with its head oriented horizontal to the ground. Panel E. The vestibular apparatus and cochlear projection of related Archosaurs compared with Allosaurus. Shown is the vestibular apparatus of the alligator and a typical bird. The vestibular apparatus is composed of the semicircular canals (angular momentum) and saccules (s, orientation to gravity). The three semicircular canals are composed of the posterior (PC) and anterior (AC) canals whose relative orientation (lines) is different between birds and alligators. Similarly the horizontal canal (pink) is a sweeping structure in birds (head movement in multiple planes) but a hooked structure in the alligator (to accommodate rapid side-to-side head motions). Allosaurus exhibits the posterior-anterior canal geometry and hooked horizontal canal that is most similar to that of an alligator. Similarly, sensory hair cells that initiate the detection of sound are located in the cochlear projection (CP, marked black line). Birds have a longer cochlear projection and consequently greater sensitivity to a broad range of frequencies relative to alligators that are limited to hearing sounds predominantly in the range of lower frequency. Allosaurus again resembles the alligator and lacks the cochlear elaboration of birds. Images adapted from Rogers [1998, 1999]. Panel F. An illustration comparing the brain and its orientation within the endocranial space between Archosaurian relatives including the proposed structure for Allosaurus derived from spiral CT scans shown in Panels B and C. The brain is positioned into the endocranial space (light gray) that is limited by the confines of the skull (dark blue lines). In alligators the brain is suspended within the endocranial space whereas in birds the brain the entire endocranial cavity, which is also true of birds that do not fly (see Rogers, [1998, 1999] for discussion). To the right is an enlarged image of the brain showing the cerebral hemisphere (CH) in light green and olfactory projections and bulb (OB) in yellow that enter the cerebral hemisphere (CH) at its most rostral aspect. The cerebellum (Cb) is blue, the diencephalon is in darker green (note the ophthalmic nerve, on) and the medulla/brain stem is gray (BS). Notable is the shift in the dependence from larger sensory structures (alligator olfactories) and poorly elaborated telencephalon [sensory information processing] to reduced primary sensory input and increased elaboration of the bird telencephalon. Reconstruction of the brain-endocranial space from the Allosaurus endocast reveals (compare these images with CT rendered images in Panel B) the brain is suspended in the brain case surrounded by supporting tissues and reveals a shape and relative proportionality of primary sensory structures to poorly elaborated telencephalon resembling that of modern alligators.

The reconstruction of dinosaur behavior usually begins with an examination of their relationship with extant species. An important contributor to this understanding began with the discovery of Archaeopteryx, a half-bird, half-reptile like creature that was championed by T.H. Huxley as a critical ‘missing-link’ organism, a discovery that was also of importance for supporting Darwin's concept of the descent of species through natural selection. However, as recently as two decades ago, a popular view was that Archaeopteryx descendants, the birds, were close relatives of mammals rather than the cold-blooded and behaviorally less complicated reptiles and alligators [Gardiner, 1982]. Phylogenetic verdicts of this type led to many proposals regarding dinosaur behavior that were based on the behaviors of mammals including such popular proposals that carnivorous dinosaur behavior could be viewed in terms of the predatory-prey relationships of modern mammals [e.g., Bakker, 1986]. Recent evidence accumulated through cladistic-based reanalysis of new discoveries in genetics and functional and comparative neuroanatomy has occasioned the need to revise these ideas. Today, birds and alligators are acknowledged to share a common ancestry with dinosaurs (Fig. 1b) whereas mammals diverged from this lineage long before (e.g., Gauthier et al. [1988]; Parrish [1997]). Further, there seems little doubt that birds arose from a specific subgroup of dinosaurs termed theropods [Forster et al., 1998] that include such notable members as the Allosaurids and Tyrannosaurids. Consequently, comparing Allosaurus with their closest living relatives, alligators and birds, is likely to be more insightful than the frequently more popular comparisons made with predatory mammals whose lineage, evolution, and brain structure differ in many important ways.

The clarification of this genealogy also hints that a substantial reconsideration of dinosaur behavior is warranted. This is particularly true since much of comparative paleoneurology was born in a time when scientists wrestled with basic concepts including social biases regarding our place within the diversity of surrounding life forms. The inevitable consequence was the use of a terminology that can suggest misleading and often anthropomorphic connections. For example, the use of brain size was often extrapolated to infer the complexity of imprecisely defined ‘behaviors’ such as intelligence, whose definition and measurement varies considerably even among members of the same species. This influence is witnessed in the work of early examiners of endocasts such as O.C. Marsh who was largely responsible for the concept that the small brain of dinosaurs implied that they were rather stupid, a not entirely uncommon belief even today. In fact, one of the earliest of Marsh's abundant finds was named Morosaurus [roughly translated; ‘stupid lizard’]. Although it is now clear that phylogenetically related groups share similar brain to body size ratios [encephalization quotient or EQ] and that brain size is not a straightforward predictor of subjective measures such as ‘intelligence’ [Jerison, 1973], the views of Marsh and his contemporaries were in effect institutionalized despite the fact that the brain of dinosaurs provided sufficient behavioral complexity to afford them collective domination of vertebrate life for roughly 100 million years.

Organisms win evolutionary success through harmonizing physical and behavioral adaptations to meet changing environmental challenges. The fossil record of Allosaurus has offered considerable physical evidence from which many of its physical abilities and disease susceptibilities have been ascertained. But what we know about the behavior of these animals has been a subject of conjecture, although impressions of the cranial cavity that once housed the brain do indeed exist for this and many other dinosaurs [Hopson, 1979]. The completely ossified cranial cavity of dinosaurs is well suited to form accurate endocranial casts [endocasts] that form when infiltrate of the cranial space becomes fossilized [Hopson, 1979]. An Allosaurus endocast, housed at the Utah Museum of Natural History (Fig. 2a) is an outstanding example of the detail that can be preserved when exceptional conditions preserve a fossil that is removed from the skull and matrix rock with great care [Madsen, 1976]. In this endocast, in addition to the preservation of cranial nerve and blood vessel location [identifiable on many endocasts; Hopson, 1979], also present are the vestibular structures that surround the flocculus (a projection from the cerebellum), the cochlear extension, and blood vessels on the endocast surface consistent with dura [Madsen, 1976; Rogers, 1998, 1999]. Although it is tempting to extrapolate the shape of the endocast to that of the brain it once housed, whether or not this can be done with accuracy depends largely on whether or not Allosaurus was more closely related to its living extant relatives, the alligators or birds. This is because the skull shape of amphibians, reptiles, and alligators is determined more by the shape and volume of sense organs such as eyes and olfactory processes (Figs. 1c, 2e) than by its brain [Jerison, 1973; Hopson, 1979]. In fact, the brain of these species may fill as little as 50% of the endocranial space while the rest harbors dura, cartilage, and blood vessels (Fig. 2e]. Hence, in these organisms the brain/endocast relationship is not always useful for studying brain shape and in some cases can actually be misleading as in certain fish whose posterior endocast cavity suggests a much larger cerebellum than is present. In contrast, the sensory organs of birds (and mammals) are reduced in size and the brain fills most of the endocranial space resulting in endocasts that are often very accurate impressions of the brain. Because of the exceptional preservation of the Allosaurus endocast, the use of spiral computed topography (spiral CT) scanning was used successfully to reveal endocast internal structure and assist in establishing the brain/endocranial cavity relationship. As reported previously [Rogers, 1998, 1999] and shown in Figure 2b,c using new images that were collected using the latest generation GE Light Speed Plus spiral CT-scanner, the relationship revealed for Allosaurus was that of alligators, and was quite distinct from that of birds (or mammals). This is because internal structures consistent with blood vessels and possibly the brain itself was seen to be suspended evenly within the endocast to create a configuration not unlike that of CT scans of modern alligators [see Rogers, 1998 for detailed discussion]. Consequently, the first impression of the relationship of the brain with its extant relatives was that of seeing Allosaurus in terms of a rather large alligator. What next?

The reconstruction of behavior from neuroanatomy, especially in an extinct species, requires the integration of numerous aspects of the animal's history in addition to the information gleaned from its endocast. This includes new and evolving views of how brain development and evolution are controlled by genes and how new imaging methods can visualize behaviors as they transpire in the brain's computational network. However, central to the premise that behaviors of extinct creatures can be reconstructed from comparisons with extant relatives is the consideration of the remarkable actuality that all vertebrates share a single and basic body/brain architecture that has been elaborated and specialized by the forces of evolution to afford greater sensory awareness through dorsally-enriched ascending (afferent) sensory-related functions and concurrently more ventrally located descending motor and autonomic (efferent) systems to adapt behaviors to match precisely the demands of the environment.

The expanding concepts of how genes control the development of distinct morphologies and how these conserved anatomical arrangements can then be used to infer behaviors establishes a basis for interspecies comparisons as well as those with extinct species. Notable in establishing the basis for such a comparison is the need to recognize that as the size of the brain is progressively increased, as exemplified by the dramatic increase in the forebrain of primates relative to reptiles, there is also the retention of common structures such as the hindbrain, cerebellum and the placement and function of cranial nerves and associated brain sensory structures. Two major mechanisms are recognized to contribute to this basic scheme of brain enlargement. First, there is a strong tendency for rostral (anterior) expansion (Fig. 1c). This appears to be the result of the function of genes (e.g., Hox genes) that exert their control in development on the overall rostral-caudal (anterior-posterior) body plan through regulating the sequential addition of segments that ultimately acquire differentiated functions. When a gene that regulates segmentation is duplicated, or other genes acquire similar interactive functions, additional segments can be added during development that have the potential of being retained in their offspring if the new material is successfully integrated with pre-existing structures and if it affords expanded or complimentary behaviors of benefit to the organism.

Second, there is lateral enlargement of the brain that is related to modifying regional cell number (i.e., complexity), a process termed elaboration (see Fig. 1c and Butler and Hodos [1996]). The sequential addition of segments to pre-existing structures and their modification through elaboration has the practical consequence of leaving in place pre-existing structures to retain time-tested interbrain relationships while diversifying and specializing others. Hence, as the size and complexity of the brain grows there are increased neuronal connections between and among areas formerly specialized to specific tasks. This, in essence, creates the potential for increased interaction between brain areas to allow for increased sensory interpretation and behavioral responses while at the same time decentralizing the location in the brain where these events emanate. This is particularly evident in mammals where large elaboration of the telencephalon (the neocortex, Fig. 1c) results in the integration of many sensory and motor functions to create precision and sophistication in behavioral responses that are not always readily predicted from comparative anatomy. Another way to appreciate this issue is to place a model of a brain from a chimpanzee and a human in front of you and guess which one uses a straw to collect termites for a meal while the other sends and then manipulates robotic rovers on Mars. These considerations lead to the necessity of defining parameters carefully when attempting to establish the relationship between brain and behavior among related species.

Since the brain changes through systematic gene-based modifications among related species, the possibility of inferring key relationships between these organisms at the level of genes, development, brain morphology and behavior appears encouraging, and that while this approach may not always be effective for establishing what organisms can do, they do provide insights into the likely limits of behavioral complexity and flexibility of the extinct species. In this context, while similarities between Archosaurs and mammals are inevitable as they share a common genetic linage it is nevertheless a distant common ancestry with fundamental differences that emerged as time separated them and the processes of brain development and elaboration became specialized to the niches occupied by each group. Therefore, in this essay I begin with an examination of the brain at the level of common mechanisms of sensory acquisition where the shape and function of these organs are imposed by the physical laws governing their function, such as detecting light or vibration. The next level of analysis describes how sensory information is integrated through brain structures that are genetically, developmentally and morphologically related. Finally, how these become progressively more integrated as the behaviors shift from those that respond almost immediately to sensory stimulation versus those that are imparted through a more complex process that involves interactive interpretation of sensory input by multiple brain regions before a behavior is initiated. Based on the presence of these key brain structures and how the brain is elaborated, comparing Allosaurus paleoneurology with the behaviors in common with its extant cousins affords a basis from which to reconstruct the likely behaviors of the animal.

A key adaptation in vertebrate history occurred during the conversion from a stationary to mobile life, a process that required a perception of orientation in space relative to gravity [Henn, 1988]. The solution was the vestibular apparatus; sensory structures that have evolved through strict adherence to the laws and limitations of basic fluid dynamics [Jones, 1974; Lowenstein, 1974; Wersäll and Bagger-Sjöbäck, 1974; Wever, 1978], and function to convert changes in fluid motion into precise signals that the brain can interpret as movement of the body in space. Hence, physical laws impose restrictions on canal diameter, size and relative geometry of the vestibular apparatus to assure that key co-variants such as fluid viscosity and/or efficiency of signal transduction are achieved. Consequently, despite the dramatic differences in the relative size of animals (e.g., whales to mice), these structures differ little in overall size. What does vary is relative canal geometry (Fig. 2e) and this is optimized to facilitate specialized behaviors. For example, in birds and mammals the vestibular canals acquire a roughly orthogonal relationship with the horizontal canal being further molded into the form of a sweeping arc. In contrast, reptiles and alligators harbor vestibular canals in a triangular arrangement where the horizontal canal is ‘hooked’ (Fig. 2e; Lowenstein [1974], Wersäll and Bagger-Sjöbäck [1974]). These differences facilitate certain body movements and limit others. For example, the hook in the horizontal canal of alligators allows the rapid side-to-side head movements during dismemberment of prey without ensuing dizziness [Rogers, 1998, 1999]. In contrast, the orthogonal canal geometry and sweeping horizontal canal of birds and mammals is better suited to a life-style that requires positioning of the body and head to facilitate functions in multiple planes of its three dimensional world [Dickman, 1996]. Fortunately, because of their bony structure, the vestibular apparatus for many dinosaurs has been retained in the fossil record [Hopson, 1979] including the Allosaurus endocast from the Cleveland-Lloyd quarry. As shown in Figure 2e (also see Rogers, 1998, 1999]), the shape of the Allosaurus vestibular apparatus is remarkable similar to that of alligators. Therefore, the adaptations and limitations imposed by the rules of harnessing of fluids for biological sensation assure us that Allosaurus, like alligators, held its head level (Fig. 2d), while living its life on a plane, relatively unaware of what might be above it, and was well designed to thrash its head rapidly side-to-side during the dismemberment of carrion or prey.

This is not the only reason an animal might thrash its head rapidly in the horizontal plane. For example fossil evidence suggests that Tyrannosaurus, whose endocasts appears to be similar to that of Allosaurus [Osborne, 1905; Brochu, 2000], may have engaged in the occasional active pursuit of prey in addition to the search for carrion. In fact, evidence of violent clashes between the Tyrannosaurus and heavily armored herbivores such as Triceratops includes findings that the herbivores were not always the victim of such encounters [Erickson and Olson, 1996]. In any case, if Tyrannosaurus regularly participated in such confrontations, they would also require skills such as rapid body movements to change direction and the more subtle and equally important abilities to persevere in task, a behavior unlikely to be necessary on a diet restricted to carrion. For the exceptionally large Tyrannosaurus accomplishing rapid changes in the momentum of its considerable mass was no small feat [Carrier et al., 2001] especially if success required rapidly moving its head to grab its prey from behind. Rapid changes in rotational inertia and the coincident carefully coordinated horizontal head movements also require an aptitude for balance during such pivots without becoming dizzy. Such movements are consistent with the configuration of the hooked horizontal canal and suggestive that the same evolutionary modifications that made possible the shredding of carrion could also accommodated additional novel behaviors while imposing restrictions on the acquisition of others.

The basic principle of the vestibular apparatus also extends to hearing where air vibrations are translated to the fluid-based movements of sensory hair cells housed within an extension of the ventral portion of the vestibular apparatus termed the cochlear duct (Fig. 2e). The length of the cochlear duct correlates with the ability to distinguish sounds over a greater range of intensity and frequency. Amphibians and reptiles, which interpret sound in part through skeletal vibration of the jawbones, display only a small specialization of this structure at its base termed the saccule [Lowenstein, 1974; Wersäll and Bagger-Sjöbäck, 1974; Wever, 1978]. In extant Archosaurs (and us) the inner ear is freed from the jaw and fluid displacement in the cochlear duct transmits the majority of sound, although a visit to the dentist reminds us that this pathway for receiving sound is not lost entirely. The ability to sense sound from vibrations of the air has facilitated substantial complexity in the cochlear projection. In alligators this projection is relatively short assuring that only sounds of low frequency or very high intensity would be heard, whereas in birds the cochlear duct is extended to provide greater numbers of auditory hair cells and in turn the increased auditory range required to communicate and receive their lovely songs (Fig. 2e). This projection in Allosaurus reveals that it was best at hearing in lower frequencies and was not well equipped to hear subtle sounds especially of higher pitch (Fig. 2e; Rogers, 1998, 1999]).

The size of the cochlear projection may not necessarily be the whole story. In our ears sound is amplified by three small bones derived from the jaw (the malleus, incus, and stapes) that connect the eardrum to the oval window of the cochlear projection. These bones amplify air displacement to increase the amount of fluid displacement and hence modulate sensitivity to volume and pitch. Further, each ear is separated by the intervening skull. This provides a temporal buffer and apportions the sound intensity between ears that together are used to by the brain to distinguish location and probable distance computationally.

Birds and alligators use a somewhat different strategy. The hearing apparatus requires only a single very thin bone (the columella, homologous to the stapes). In fact, the columella of Allosaurus and that of Tyrannosaurus could be several centimeters long [Colbert and Ostrom, 1958]. Therefore, while sound is amplified, it does not reach the level observed in mammals with three inner ear bones. In addition, the eardrums of these species are not separated entirely by intervening bone but rather they are linked by a passage through the skull that permits the almost simultaneous reception of sound by both ears. Recent studies of how crocodiles detect sound and locate it in space [Soares, 2002] suggest that these organisms also use a very different strategy than mammals. In this case, a portion of the function of the cochlear region has been dispersed to dedicated sensory organs that line the upper surface of the face [termed hypertrophied nerve system] for detecting vibration. Since the vibration from a prey item is detected along these cells asymmetrically, this effectively distributes cochlear function into structures that specialize in sound recognition and ‘prey’ recognition. Consequently, while alligators hear low frequency sound and use it sparingly for communication, the hypertrophied nerve system is well-tuned to detect asymmetric vibration and assure that a potential prey item, such as a golfer in search of a lost ball, is accurately located before initiating a lunge. There is little evidence (to my knowledge) for the presence of a hypertrophied nerve system in Allosaurus, nor would it be easy to detect since its preservation in the fossil record would be limited and easily overlooked [Soares, 2002]. However, if Allosaurus sensory structures where elaborated through dissemination and specialization, this would modify substantially our ideas about the capabilities of this creature in how the nervous system could in effect be elaborated without such modifications being reflected by the brain itself. Given these caveats, the overall configuration of the Allosaurus nervous system suggest that this animal was unlikely to have used complex vocal embellishments to communicate and it was not likely to have sufficient elaboration for anything other than an immediate response to sensory input. Consequently it would have lacked the structural specializations related to facilitating the more complex and flexible behaviors of its extant avian relatives or those of its more distant mammalian relatives.

Not all sensory input is equally constrained as those related to balance and hearing. This takes us to the next level of evolutionary boundaries on brain development and expansion. Most notable in this category are the sensory and motor nerve relationships first recognized by Ramon Y Cajal over 100 years ago that distribute sensory input efficiently to centers of the brain where they can be integrated into coordinated motor responses. Allosaurus is clearly among the creatures subject to these basic organizational restrictions as reflected by the retention of a basic vertebrate body plan that is recognized immediately in the endocast by the presence of all major cranial motor, sensory and even circulatory structures (Fig. 2a; Hopson [1979], Rogers [1998]].

The requirement to conserve these structures through timing gene interactions during development without introducing detrimental effects is a point from which similarities in anatomy and function can be expanded. For example, the Hox genes were among the first master-switch genes discovered whose overlapping hierarchy of tightly regulated sequential expression contributes to proper segmentation of the embryo during developmental and eventual tissue patterning (e.g., Puelles and Rubenstein [1993], Butler and Hodos [1996], Schilling and Knight [2001], Carpenter [2002], Santagati and Rijli [2003]). These genes and the topographic organization of the segments whose development they regulate are similar in all vertebrates and they provide one example of how anterior-posterior structure function-interactions are determined and retained in evolution. Further, the interdependency of developmental programs that lead to the formation of seemingly independent structures such as fore-limbs and hind-brain can result in the imposition of developmental constraints that become progressively more difficult to modify as evolution builds on these pre-existing structures. Elaboration can nevertheless be accommodated by varying those regulatory processes closely associated with determining cell number. An example is the HoxD gene complex that contributes to patterning the development of arms, digits and several key aspects of the hindbrain. Here the genes are subject to regulatory modifications that do not qualitatively alter homologous structures, but do modify them quantitatively. This is seen in bats where the forearms are extended relative to land-based mammals to become wings without alteration to the basic hindbrain neuroanatomy. Similarly, ancient Archosaurs and related groups almost certainly shared similar capacities to selectively modify Hox and other developmental gene functions as suggested by the changes observed in the forelimbs of theropods, whose forearms were shortened in Tyrannosaurus to effective uselessness, but lengthened in others such as Archaeopteryx to become primordial wings.

How does this apply to Allosaurus behavior? One sensory system that is primitive and anatomically highly conserved, but can vary considerably in size and contribution to behavior, is olfaction (Fig. 1c). Olfaction is among the earliest sensory functions to appear in evolution where chemical signals are converted to electrical signals that are processed by intrinsic circuits of the olfactory bulb before again being transmitted directly (by-passing thalamic relays in vertebrates) to multiple basal forebrain structures including olfactory and piriform cortex and limbic structures for additional signal processing and interpretation. This scheme has important implications on how this basic morphological configuration, within the context of extant species, sets limits on behavior. In reptiles and alligators, for example, the primary olfactory bulb comprises a substantial portion of the forebrain (Figs. 1c, 2f). Most neurons at this level are dedicated directly to the task of chemical response such as smell and little forebrain elaboration is present to process this information further. Consequently, sensory signals are processed mostly at the level of the sensory organ and the results of that processing are sent to the limited forebrain where they impart behaviors bordering on a CNS reflex arc. This means that if an object smells like food, it is likely swallowed, which can result in significant feeding errors. However, feeding errors in alligators are tolerated because of the relatively low energy demands of ectothermic (cold-blooded) metabolism. In contrast, the demands of endothermic metabolism in birds (and mammals) are incompatible with an equivalent lack of sensory discrimination. This places greater demand on processing sensory information before responding to it, and coincides with greater forebrain elaboration and a diminished olfactory bulb size (compared to that of alligators) resulting in a greater correspondence between the endocast and brain shape (Fig. 2f). Hence, while some key reflex-like connections remain (a bad smell can ‘turn your stomach’), increased cortical processing conveys more proficient behaviors to assure efficient discriminatory feeding thereby improving the diet to meet the demands of exothermic metabolism. As noted above, the endocast of Allosaurus harbors large primary olfactory bulbs and a relatively poorly elaborated telencephalon (Fig. 2f) suggesting strongly that their related behaviors more closely resembled those of alligators and not those of birds. Therefore, in addition to a vestibular system designed for thrashing of the head, Allosaurus most certainly relied heavily on its sense of smell and thought little before thrashing and swallowing essentially anything suspected to be a food item.

Using comparative neuroanatomy to extrapolate to the possible range of complex behaviors of Allosaurus requires a very careful assessment of the relationship between structural and functional homology among brain regions. Thus a structure is ‘homologous because of what it is, not because of what it does’ or contrariwise the presence of an equivalent structure is not necessarily evidence of an equivalent function. In the absence of direct evidence, part of the solution to this conundrum of extrapolating homology-function relationships in extinct species returns us to the time-tested anatomical comparisons with structures of extant species with the twist that many of these comparisons can be made with greater confidence using the findings emerging from functional comparative neuroanatomy. An example is the structure of Archosaurian descendants termed the dorsal ventricular ridge (DVR; Karten [1991]) that, with its underlying pallium, serves a function similar to that of the mammalian neocortex. However, the cellular architecture of the DVR and associated structures is more similar to that of the mammalian striatum, which is a non-laminated neuronal array that produces the equivalent of a ‘switchboard’-like input and output system. This is in contrast to the laminated structures of the cortex where neurons of different functions are arrayed in distinct layers that receive, integrate, modulate and finally disseminate information in hierarchal patterns. In mammals the non-laminated striatum is particularly well-suited to refine repetitive fine motor skills (often through habituation), while auditory afferent relays from the thalamus terminate in the laminated structures of the cortex where auditory and vocal skills are integrated and processed to allow for flexibility in responses. In contrast afferents from thalamic auditory relays of Archosaurs are connected to the non-laminar structures of the DVR and underlying pallium (for extensive reviews of this developing area or study see Karten [1991], Butler and Hodos [1996], Striedter [1997], Aboitiz et al. [2002], Molnar and Butler [2002], Soares [2002]). This has the effect of auditory information and vocalization being processed differently than in mammals and assures that the memorization and repetition of species-specific songs are effectively behavioral mimicry (e.g., parrot replication of human speech). This limits the plasticity of applying and modifying these sounds to unique situations as can occur when the behavior emanates from laminated structures. In crocodiles and alligators, even less of this region is dedicated to sound recognition and more to the interpretation of vibrations from the hypertrophied nervous system [Soares, 2002]. Since Allosaurus was a member of the Archosaurian group, it follows that the forebrain contained the DVR configuration. Placing this in context with the vestibular shape, limited cochlear extension and relatively poor elaboration of the telencephalon relative to olfactory sensory structures, it would seem apparent that the related behaviors of Allosaurus were highly repetitive, lacked flexibility and the role of sound was relatively small except perhaps for the low-frequency but high volume methods of long-range communication [e.g., roaring] important to ritualized behaviors.

But, could there have been unique or more complex behaviors that originated in multiple brain centers [Halpern, 2001; Marinkovic, 2004] that in Allosaurus could not be predicted by comparisons with extant relatives? For example, the remains of dinosaurs have been found hovering over their nests recording the last moments of this animal's nesting behavior [Horner and Weishampel, 1988; Geist and Jones, 1996; Kellner, 1996; Ruben et al., 1999; Ruben et al., 2003]. Hence, the protection and care of future offspring is a behavior that is more complex than that of reptiles who lay eggs in numbers sufficient for future generations to make it on their own. Based solely on neuroanatomical comparisons, altruistic behaviors are not easily predicted from the very similar brain shape of reptiles with those of the Archosaurian lineage [Geist and Jones, 1996]. Unfortunately such limitations in predicting rudimentary (but nonetheless complex) behaviors have in part contributed to suggestions that these creatures harbored significantly greater behavioral capabilities than is likely to have been possible. The fact remains that the brain of Allosaurus was not highly elaborated nor does it exhibit any unusual complexities beyond that of the alligator whose brain controls behaviors that are specialized and decidedly regimented.

Further, the regions of the brain responsible for behaviors such as intricate emotions or logic do not exist or are poorly developed relative to mammals in essentially all Archosaurs, as are others regions such as the hippocampus where memory consolidation occurs and many key behaviors such as vigilance originate [see Butler and Hodos, 1996]. This would have been the case for Allosaurus. Therefore, while certain behaviors appear complex in their execution (such as the ritualized mating encounters of alligators; Vliet [1989]), the origins of these behaviors are deeply rooted in the gene-based morphological relationships that are poorly amenable to modification since the pressures of natural selection tend to reward successful repetitive performances rather than clever individualized improvisations. Consequently, comparing alligators with Allosaurus suggests that these extinct creatures would also have engaged in highly ritualized behaviors including protecting their nests and caring for their young. Nevertheless, it was equally probable that the evolutionary success of this creature did not preclude an occasional digression or miscue by an aberrant sensory input resulting in its eating the young of adjacent nests, or possibly even its own.

But times changed and for Allosaurus these changes exceeded its ability to adapt behaviorally and they became extinct. Yet Tyrannosaurus, whose endocasts resembles those of Allosaurus [Osborne, 1905; Hopson, 1979; Rogers, 1999] but possess olfactory processes of even greater relative size to the telencephalon [Osborne, 1905; Brochu, 2000], emerged as a dominant carnivorous dinosaur of the Cretaceous period. Curiously, such modifications might suggest that Tyrannosaurs was behaviorally even more dependent on sensory stimulation such as smell, and possibly behaviorally even more rigid than Allosaurus despite the novel challenges that these giant carnivores faced. This returns us to the concept that at the present time analysis of the brain suggests behaviors in broad strokes rather than specific detail. So as for alligators or even birds, who repeat key sets of behaviors in rapid succession until something accomplishes the task (as might occur if confronted with an unknown situation such as capture), they all share in the constraints and parallels imposed on them by virtue of their inherited neurological configuration. Consequently, as relationships between genetics, morphology and behaviors become better resolved, we can look forward to equally improved confidence in our understanding of the intentions of Allosaurus as it moved toward what would become its fate in the Cleveland-Lloyd quarry some 130 million years ago.

In summary, validating how Allosaurus viewed its world will require that we first acquire a better understanding of the fossil record in the context of a much more difficult challenge to understand the relationship between genes, neuroanatomy and behavior. Not unlike earlier scientists who tackled the overwhelming uncertainties of resolving physical laws governing the world, today the substance of elementary school education, we face the task of comprehending how the laws of genetics govern the transformations of morphology into the structures that ultimately govern behavioral possibilities and constraints. This daunting task is tempered by the recognition by Francis Bacon that time is on our side when resolving these seemingly complex and overwhelming issues. Or so we hope. The brain of Allosaurus received sensory information and dictated behaviors appropriate for successful domination of its environment for about 40 million years. Yet slow environmental changes punctuated by occasional more rapid or even cataclysmic events [Gould, 1990] ultimately challenged Allosaurus beyond its abilities to adapt whereas alligators, birds, and mammals flourished. Therefore, as we identify similarities between Allosaurus and its extant cousins, we are also reminded that it is the differences in behavior and function of the organisms that lead to the success of one group and the failure of others. What modern science continues to offer is an ever-expanding concept of brain-behavior relationships whose extrapolation to this group of extinct species offers an ever better view of their life and times during a truly enigmatic period on this planet.