The richness of the results from locomotor behavior in rodents in terms of elucidating the underlying neural mechanisms has not been paralleled by similar studies in humans. Thus, to date a comparable laboratory-based, multivariate assessment of locomotor activity has not been extended to human populations. Some earlier studies have measured motor activity in psychiatric populations with the use of a wrist or leg accelerometer (Teicher 1995
, Teicher et al. 1986
, Wolff et al. 1985
). The vast majority of studies on hyperactivity, however, have been limited to observer-rated and self-report scales. As discussed above, such scales may not be optimal in detecting potentially subtle alterations in activity levels nor are they informative about qualitative aspects of hyperactivity that may distinguish certain psychiatric populations from others. For example, several diagnostic groups may present with symptoms of hyperactivity but they may be qualitatively distinct and thus reflect different underlying neural circuitry abnormalities.
In response to the need for models of mania and following our work in rodents, we developed the human Behavioral Pattern Monitor(human BPM), as an analog of the rodent BPM and a method with which to sensitively quantify the characteristics of human hyperactive and exploratory behavior. Unlike other translational paradigms, the human BPM reflects a “reverse-translational” approach using the existing rich animal literature to inform its development. This approach contrasts with other well-established translational paradigms such as prepulse inhibition of the startle response, an index of sensorimotor gating. The assessment of prepulse inhibition in psychiatric patients was first developed in humans using the startle eyeblink response (Braff et al. 1995
). It was subsequently extended to rats (Geyer et al. 2001
) and then mice (Geyer et al. 2002
), using the whole-body startle response. In contrast, the human BPM represents an evolution in the opposite direction, where a paradigm originally developed in rodents is mimicked in humans. The implementation of prepulse inhibition and the human BPM are examples of the bidirectionality of translational science, where human studies inform animal research, and vice-versa.
The human BPM takes place in a 9' by 14' room that the human participant has not been exposed to and therefore is, like the rodent BPM, a novel and unfamiliar environment. As detailed below, similar to the rodent BPM, multiple measures of motor activity can be collected, including spatial d, entropy h, transitions, distance traveled, and others. Along the walls of the room, dispersed evenly on items of furniture, are ten small objects. These objects were chosen using the criteria that they be safe, colorful, tactile, and manipulable and therefore invite human exploration. The objects provide an analog of the exploratory holes in the walls and floor of the rodent BPM chambers. Participants are directed into the room with little instruction or direction and are asked to wait for the experimenter to return. The human BPM session has been fifteen minutes long in our studies to date.
Data in the human BPM are gathered using three sources of measurement: 1) collection of physiologic data, namely motor activity of the subject's torso, using an accelerometer embedded in an ambulatory monitoring device that the participant wears; 2) x-y coordinates of the subject's spatial location in the BPM, extracted from digital video recording; and 3) experimenter ratings of exploratory activity, obtained by carefully scoring the video recording of the BPM session and measuring events such as interactions with objects. These three sources of measurement capture different qualitative aspects of motor and exploratory behavior, and yet may also be intercorrelated in the case of certain types of behavior, as will be illustrated below. In fact it is hypothesized that similarly to the rat and mouse BPM, three main independent factors will emerge, describing activity (accelerometry, transitions from one region to a neighboring region), exploratory behavior (interaction with objects), and sequential organization of behavior (spatial dand spatial CV).
In the human BPM, one measure of the amount of activity is quantified with an accelerometer, which is embedded in a wearable ambulatory monitoring device. The LifeShirt System (Vivometrics 2002
) is an ambulatory, multi-sensor, continuous monitoring system that collects objective physiologic data through various sensors, including respiratory inductive plethysmography bands, which measure pulmonary function, electrical activity of the myocardium via a 3-lead EKG, and activity/posture via a two-axis accelerometer. The sensor array of the LifeShirt System is embedded in a sleeveless undergarment. For measurement of activity level, a two-axis accelerometer is placed onto the shirt over the sternum, and the rectified and integrated accelerometer signal is used to detect periods of physical activity and rest. An on-board PDA continuously encrypts and stores the patient's activity and postural physiologic data on a compact flash memory card. Accelerometry data are sampled at 10 Hz and stored numerically in digital units. Thus, one measure of the amount of motor activity is obtained by averaging acceleration values over the three five-minute intervals of the human BPM session. Exemplars of the acceleration values derived from individual subjects are provided in .
To obtain additional measures of the quantity and patterns of motor activity and exploratory behavior, the room is also equipped with a camera and fish-eye lens system hidden in a ceiling vent. The images from the camera are stored in digital format on a computer in the adjacent room, with a frequency of 30 frames per second. The digital videos of subject's activity in the human BPM are subjected to frame-by-frame analysis with proprietary software (Clever Systems, Inc. 1999), which generates xand y-coordinates of the subject's successive locations. Because the software specifically tracks the blue LifeShirt vest, the coordinate positions reflect the position of the upper part of the subject's torso. At present, these x-y coordinates are used to plot the path of the subject and to count time spent and transitions between nine arbitrarily defined regions of the human BPM. These regions are analogous to our definition of nine areas of the rodent BPM (Geyer et al. 1986
), namely the four corners, four walls, and the center. Delineation of these regions allows us to obtain a distribution of amount of time spent in each region as well as to measure the number of transitions, defined identically to the rodent work as movement from one region to an adjacent one. As we gain further experience with data generated using this system, alternative definitions of regions will no doubt be found to be more relevant to the behavior of humans in this environment than are the regions defined previously for use in rodents. In any event, transitions between regions and dwell times within specific regions can serve as additional measures to describe different aspects of locomotor activity and to complement the accelerometry data. In addition, the digitized video images enable detailed assessments of the subject's interactions with the 10 objects placed in the room, in analogy to the rodent's investigatory behavior directed toward the 10 holes placed in the walls and floors of the rodent BPM chambers.
The continuous, high-frequency sampling of motor activity data also allows us to calculate dynamical entropy, h, which is comparable to the entropy measure mentioned above in the context of our animal studies. Dynamical entropy quantifies the predictability of a given level of activity based upon preceding patterns of activity. This acceleration-derived entropy measure captures, to our knowledge, a unique feature of human locomotor behavior, i.e. how sequences of acceleration events are organized in time. More importantly, we have already been able to use it to derive entropy “signatures” for specific and distinctive patterns of motor behavior. For example, in initial standardization studies, we generated average entropy values for motor behaviors that subjects exhibited in response to audio-taped instructions, e.g. walking, sitting, standing motionless, or exploring an environment. These entropy values were then used to generate mathematical probabilities that new subjects, uninstructed in the human BPM, were engaging in those motor behaviors. illustrates that videotape ratings of a subject's walking behavior corresponded precisely with the entropy-derived mathematical “signature” of walking. Given that dynamical entropy h can characterize disordered movement as well as perseverative movement as described above in our animal studies, it holds much promise as a potentially informative measure of how human motor behavior is organized across time.
Observed walking and entropy-derived probability
A measure of sequential organization that can be derived from the human BPM data and is completely analogous to our measurement of the organization of rodent locomotor data is the spatial scaling exponent, d
, which, as in the case of the rodent work reviewed above, describes the geometric pattern of a subject's movement in the exploratory environment. Spatial d
is derived in a near-identical manner to the rodent BPM (Paulus et al. 1990
, Ralph et al. 2001
). The series of x-y-coordinates derived from the digitized video images describes the spatial patterns of the subject's location and is used to calculate spatial d
over specified time blocks.
Although the human BPM is in some respects a novel effort to develop a parallel to the animal open field paradigm, laboratory-based exploratory environments for humans have been reported previously. An early study examined the exploratory behavior of infants in a novel environment (Rheingold and Eckerman, 1969
). Similarly, (Pierce and Courchesne 2001
) quantified exploratory behavior in autistic children by rating videotapes of an eight-minute session where subjects were placed in a room with colorful and interactive objects. Ratings of decreased exploration were correlated with MRI-based measures of altered brain volumes in children with autism, suggesting that this exploratory paradigm was useful in detecting behavioral deficits that are associated with brain dysfunction.
Current Applications and Future Directions
The human BPM is one of the central measures we are using in an ongoing investigation of inhibitory deficits in bipolar mania. The original basis for developing the human BPM was to conduct parallel, cross-species studies of inhibitory problems that are features of the mania of BD and to extend the paradigm to other conditions such as schizophrenia, where multivariate assessment of motor behavior can reveal distinctive characteristics of the illness. An important aim of this study was to develop and validate rodent models of mania, which has been identified as a need in the literature (Einat 2006
). This validation would be in part based upon the potential similarities between the motor activity of manic patients in the human BPM and the corresponding rodent BPM studies of mice that have been genetically or pharmacologically manipulated to create trait or state conditions of hyperdopaminergia. Thus, as noted above, the human BPM is an example of a “reverse-translational” approach to neuroscience research: whereas most paradigms that are eventually applied to both humans and animals are first developed in humans and then modified to be tested in animal models, the human BPM is unique in that it was developed as an analog to a widely used and highly influential animal paradigm, the open field as elaborated into the rodent BPM. Thus far it shows great promise in characterizing the motor behavior of human clinical populations, both in terms of some of the more straightforward measures such as accelerometry and video ratings, as well as the more complex measures of entropy and patterns of sequential movements in space.
We are currently administering the human BPM paradigm to manic BD patients and individuals with schizophrenia who have been hospitalized on an inpatient psychiatric unit for an acute exacerbation of their illness. In we illustrate representative case examples of human BPM data for our clinical populations as well as the non-patient cohort. The x-y coordinate tracings of the manic BD patient clearly show a very high level of activity in the BPM (). Both the average acceleration and the number of transitions during the BPM session are substantially higher than those of the schizophrenia patient () or the healthy comparison subject (). In addition, the manic patient exhibits markedly more interactions with the exploratory objects. The relatively low spatial d
in the manic patient suggests that his motor behavior is characterized by long, straight movements from one area of the room to the next. This pattern of increased motor activity and increased exploratory behavior in combination with a reduced spatial d
is comparable to what we have observed in DAT KD mice (see ) and mice administered GBR12909 (see ), suggesting that DAT KD and GBR12909-treated mice may be intriguing candidates for genetic and pharmacological animal models of mania. In contrast, the schizophrenia patient exhibits very low motor activity, little exploration of objects, and a higher spatial d
, signifying restricted and localized activity. The striking difference between the manic BD and the schizophrenia patient once again highlights the importance of multivariate assessment of activity, where measurement of multiple parameters may yield distinct “signatures” of locomotor activity that characterize and differentiate these two disorders. From a diagnostic perspective, the human BPM may be able to quantitatively assess an obvious and meaningful difference between two acutely ill populations who, during acute states, are often difficult to distinguish from one another because the behavioral presentation of both patient groups is dominated by psychotic and mood symptoms (Pini et al. 2004
The potential objective, sensitive, and multivariate characterization of hyperactivity that is afforded by the human BPM offers many directions for future research. One obvious application would be to conduct pharmacological manipulations in parallel animal and human studies. For example, while the effects of stimulants on rodent motor behavior have been thoroughly characterized in the rodent BPM, studying stimulant-induced hyperactivity in the human BPM may help us further elucidate the behavioral features of an acute hyperdopaminergic state in healthy humans. Similarly, the human BPM may be useful in testing the efficacy of compounds for characterizing disorders that have hyperactivity as a central symptom. Such comparisons could include patients with BD, schizophrenia, schizoaffective disorder, and attention-deficit/hyperactivity disorder (ADHD) as well as developmental illnesses such as autism spectrum and impulse control disorders. As part of our current study on bipolar mania, we are re-testing manic subjects in the human BPM after several weeks of stabilization on psychotropic medications. We hypothesize that patients who are treated with a combination of an antipsychotic and mood stabilizer will show faster alleviation of symptoms of hyperactivity than those patients treated with a mood stabilizer alone, since the antipsychotic medications act directly as dopamine antagonists while mood stabilizers probably only indirectly modulate dopamine levels. While ours is a naturalistic study in which manic patients are not randomized to medications, an obvious future direction is to carry out randomized, controlled investigations of the effects of psychotropic compounds on hyperactivity, and to examine the time course of these effects. A possible limitation of the human BPM in longitudinal studies, however, is the potential effect of habituation, insofar as the human BPM ceases to be a totally novel environment with repeated exposures. The same problem is evident in longitudinal studies in rodents. One way to address this issue in a controlled fashion would be to design future studies where a group of subjects is first tested in a medicated state and re-tested after withdrawal from medications. These are obviously challenging studies to implement but they hold much promise for informing the field about the efficacy of psychotropic medications.
In conclusion, the human BPM is an important example of cross-fostering translational research. Given the importance of hyperactivity in many psychiatric disorders in general and in bipolar mania in particular, it is surprising that experimental approaches to measure locomotor behavior empirically in humans have not been more abundant in the literature. Our experience with locomotor behavior in rodents has shown that it is a complex phenotype that is not sufficiently characterized by quantifying only the amount of behavior. Instead, measures that quantify its temporal, spatial, and dynamic organization have proven to be valuable tools to differentiate the contributions of different neural transmitter systems on locomotor and exploratory behavior. Similarly, we predict that multivariate approaches to human locomotor and exploratory behavior will provide powerful insights into the neural bases of these behaviors and may provide new biomarkers as targets for the development of novel antimanic agents.