T-pattern detection and analysis for the discovery of hidden features of behaviour.

BACKGROUND: The behaviour of all living beings consists of hidden patterns in time; consequently, its nature and its underlying dynamics are intrinsically difficult to be perceived and detected by the unaided observer. METHOD: Such a scientific challenge calls for improved means of detection, data handling and analysis. By using a powerful and versatile technique known as T-pattern detection and analysis (TPA) it is possible to unveil hidden relationships among the behavioural events in time. RESULTS: TPA is demonstrated to be a solid and versatile tool to study the deep structure of behaviour in different experimental contexts, both in human and non human subjects. CONCLUSION: This review deepens and extends contents recently published by adding new concepts and examples concerning the applications of TPA in the study of behaviour both in human and non-human subjects.

Discovery of recurring behavioural sequences in Wistar rat social activity: Possible support to studies on Autism Spectrum Disorders.

This study was undertaken to investigate whether, in rat interactive activities, recurring sequences of behavioural events might be identified and how and to what extent each component of the pair is involved. To this aim, the multivariate temporal-pattern (t-pattern) analysis was applied to the social interactions of 9 pairs of male Wistar rats tested in open field. Interactive activities were classified into intra- and inter-subjects. Quantitative evaluations showed that intra-subject behavioural elements represented 62.37% and inter-subject ones 37.63% of the comprehensive behaviour. T-pattern analysis revealed the presence of 221 different t-patterns organized in four different categories: containing exclusively inter-subject elements; containing both inter- and intra-subject elements; consisting of rat 1 and rat 2 intra-subject elements and, finally, consisting of intra-subject elements carried out by one of the two subjects. Results show that the activity of two interacting Wistar rats is structured on the basis of several recurring temporal sequences. Moreover, social interactions appear to be expressed also by t-patterns where the behavioural elements are carried out by animals seemingly not interacting. A support of t-pattern analysis to studies on Autism Spectrum Disorders is proposed.

Temporal patterns of rat behaviour in the central platform of the elevated plus maze. Comparative analysis between male subjects of strains with different basal levels of emotionality.

We have analyzed the temporal patterns of behaviour of male rats of the Wistar and DA/Han strains on the central platform of the elevated plus maze. The ethogram encompassed 10 behavioural elements. Durations, frequencies and latencies showed quantitative differences as to walking and sniffing activities. Wistar rats displayed significantly lower latency and significantly higher durations and frequencies of walking activities. DA/Han rats showed a significant increase of sniffing duration. In addition, DA/Han rats showed a significantly higher amount of time spent in the central platform. Multivariate T-pattern analysis revealed differences in the temporal organization of behaviour of the two rat strains. DA/Han rats showed (a) higher behavioural complexity and variability and (b) a significantly higher mean number of T-patterns than Wistar rats. Taken together, T-pattern analysis of behaviour in the centre of the elevated plus maze can noticeably improve the detection of subtle features of anxiety related behaviour. We suggest that T-pattern analysis could be used as sensitive tool to test the action of anxiolytic and anxiogenic manipulations.

T-pattern analysis for the study of temporal structure of animal and human behavior: a comprehensive review.

A basic tenet in the realm of modern behavioral sciences is that behavior consists of patterns in time. For this reason, investigations of behavior deal with sequences that are not easily perceivable by the unaided observer. This problem calls for improved means of detection, data handling and analysis. This review focuses on the analysis of the temporal structure of behavior carried out by means of a multivariate approach known as T-pattern analysis. Using this technique, recurring sequences of behavioral events, usually hard to detect, can be unveiled and carefully described. T-pattern analysis has been successfully applied in the study of various aspects of human or animal behavior such as behavioral modifications in neuro-psychiatric diseases, route-tracing stereotypy in mice, interaction between human subjects and animal or artificial agents, hormonal-behavioral interactions, patterns of behavior associated with emesis and, in our laboratories, exploration and anxiety-related behaviors in rodents. After describing the theory and concepts of T-pattern analysis, this review will focus on the application of the analysis to the study of the temporal characteristics of behavior in different species from rodents to human beings. This work could represent a useful background for researchers who intend to employ such a refined multivariate approach to the study of behavior.

Multivariate temporal pattern analysis applied to the study of rat behavior in the elevated plus maze: methodological and conceptual highlights.

Aim of this article is to illustrate the application of a multivariate approach known as t-pattern analysis in the study of rat behavior in elevated plus maze. By means of this multivariate approach, significant relationships among behavioral events in the course of time can be described. Both quantitative and t-pattern analyses were utilized to analyze data obtained from fifteen male Wistar rats following a trial 1-trial 2 protocol. In trial 2, in comparison with the initial exposure, mean occurrences of behavioral elements performed in protected zones of the maze showed a significant increase counterbalanced by a significant decrease of mean occurrences of behavioral elements in unprotected zones. Multivariate t-pattern analysis, in trial 1, revealed the presence of 134 t-patterns of different composition. In trial 2, the temporal structure of behavior become more simple, being present only 32 different t-patterns. Behavioral strings and stripes (i.e. graphical representation of each t-pattern onset) of all t-patterns were presented both for trial 1 and trial 2 as well. Finally, percent distributions in the three zones of the maze show a clear-cut increase of t-patterns in closed arm and a significant reduction in the remaining zones. Results show that previous experience deeply modifies the temporal structure of rat behavior in the elevated plus maze. In addition, this article, by highlighting several conceptual, methodological and illustrative aspects on the utilization of t-pattern analysis, could represent a useful background to employ such a refined approach in the study of rat behavior in elevated plus maze.

Significant divergences between the temporal structure of the behavior in Wistar and in the spontaneously more anxious DA/Han strain of rats tested in elevated plus maze.

The aim of present research is to study the temporal structure of the behavior in two strains of rats with different basal level of emotionality. To this purpose, the temporal profile of the behavior in Wistar rat and in the spontaneously more anxious DA/Han strain was analyzed in the Elevated Plus Maze. Both quantitative and multivariate t-pattern analyses were carried out. In comparison with Wistar, DA/Han subjects showed a significant reduction of the permanence in open arm and a significant increase of the time spent in the central platform of the maze. Mean frequencies of each behavioral element showed significant modifications both in open and in closed arm. Multivariate t-pattern analyses demonstrated a very different temporal profile of behavior in the two strains: Wistar rats presented 197 t-patterns of different composition, whereas DA/Han rats only 26; as to the mean number of t-patterns, Wistar presented a value of 698.90, whereas DA/Han only 92.80. Similar clear-cut differences were detected for the mean number of t-patterns in the two arms and for the time course of such t-patterns. Present study has evidenced that the temporal organization of the bahavior in Elevated Plus Maze is differently structured in two strains of rats with different basal level of emotionality.

Temporal structure of the rat's behavior in elevated plus maze test.

Aim of the research was to evaluate, by means of quantitative and multivariate temporal pattern analyses, the behavior of Wistar rat in elevated plus maze (EPM) test. On the basis of an ethogram encompassing 24 behavioral elements, quantitative results showed that 130.14 ± 8.01 behavioral elements occurred in central platform and in closed arms (protected zones), whereas 88.62 ± 6.04 occurred in open arms (unprotected zones). Percent distribution was characterized by a prevalence of sniffing, walking and vertical exploration. Analysis of minute-by-minute duration evidenced a decrease for time spent in open arms and central platform and an increase for time spent in closed arms. As to multivariate t-pattern analysis, 126 different temporal patterns were detected. Behavioral stripes, summarizing distribution of such t-patterns along time, showed that several t-patterns were not homogeneously distributed along the test observational period: t-patterns encompassing behavioral events occurring prevalently in central platform-open arms were observed during the first minutes, whereas t-patterns structured on the basis of events occurring mainly in central platform-closed arms were detected during the last minutes. Therefore, during the observation in elevated plus maze, rat's behavior undergoes significant rearrangements of its temporal features. Present research demonstrates, for the first time, the existence of complex and significantly timed behavioral sequences in the activity of Wistar rats tested in elevated plus maze. Application of t-pattern analysis can provide useful tools to characterize the behavioral dynamics of anxiety-related rodent behavior and differentiate the effect of various anxioselective substances.

Behavioural comparison of human-animal (dog) and human-robot (AIBO) interactions.

The behavioural analysis of human-robot interactions can help in developing socially interactive robots. The current study analyzes human-robot interaction with Theme software and the corresponding pattern detection algorithm. The method is based on the analysis of the temporal structure of the interactions by detecting T-patterns in the behaviour. We have compared humans' (children and adults) play behaviour interacting either with an AIBO or a living dog puppy. The analysis based on measuring latencies and frequencies of behavioural units suggested limited differences, e.g. the latency of humans touching the dog/AIBO was similar. In addition other differences could be accounted for by the limited abilities of the robot to interact with objects. Although the number of interactive T-patterns did not significantly differ among the groups but the partner's type (whether humans were playing with dog or AIBO) had a significant effect on the structure of the patterns. Both children and adults terminated T-patterns more frequently when playing with AIBO than when playing with the dog puppy, which suggest that the robot has a limited ability to engage in temporally structured behavioural interactions with humans. As other human studies suggest that the temporal complexity of the interaction is good measure of the partner's attitude, we suggest that more attention should be paid in the future to the robots' ability to engage in cooperative interaction with humans.

Detection of temporal patterns in dog-human interaction.

A new time structure model and pattern detection procedures developed by (Magnusson, M.S., 1996. Hidden real-time patterns in intra- and inter-individual behaviour description and detection. Eur. J. Psychol. Assess. 12, 112-123; Magnusson, M.S., 2000. Discovering hidden time patterns in behaviour: T-patterns and their detection. Behav. Res. Methods, Instrum. Comput. 32, 93-110) enables us to detect complex temporal patterns in behaviour. This method has been used successfully in studying human and neuronal interactions (Anolli, L., Duncan, S. Magnusson, M.S., Riva G. (Eds.), 2005. The Hidden Structure of Interaction, IOS Press, Amsterdam). We assume that similarly to interactions between humans, cooperative and communicative interaction between dogs and humans also consist of patterns in time. We coded and analyzed a cooperative situation when the owner instructs the dog to help build a tower and complete the task. In this situation, a cooperative interaction developed spontaneously, and occurrences of hidden time patterns in behaviour can be expected. We have found such complex temporal patterns (T-patterns) in each pair during the task that cannot be detected by "standard" behaviour analysis. During cooperative interactions the dogs' and humans' behaviour becomes organized into interactive temporal patterns and that dog-human interaction is much more regular than yet has been thought. We have found that communicative behaviour units and action units can be detected in the same T-pattern during cooperative interactions. Comparing the T-patterns detected in the dog-human dyads, we have found a typical sequence emerging during the task, which was the outline of the successfully completed task. Such temporal patterns were conspicuously missing from the "randomized data" that gives additional support to the claim that interactive T-patterns do not occur by chance or arbitrarily but play a functional role during the task.

OBSERVE: a multimedia course on the observational analysis of behavior.

OBSERVE is a preliminary release of a multimedia course for teaching undergraduate and graduate students how and why to study behavior by direct observation. The instructional text and commentary and the self-test and examination materials are built around a series of exercises in which the student observes and categorizes film clips of the behavior of several different species in several different ways. Incorporation of elements from The Observer software for computer recording and video analysis implements fully computerized continuous recording. At present, the text, together with check sheets that the program generates, enables a comparison between one-zero, instantaneous, and continuous sampling of the same behavioral excerpts. Matrices are printed out for an exercise in calculating interobserver reliability. Another section supports carrying out and writing up a small observational project on human behavior in the field. Plans for the future development of OBSERVE are briefly described.

Discovering hidden time patterns in behavior: T-patterns and their detection.

This article deals with the definition and detection of particular kinds of temporal patterns in behavior, which are sometimes obvious or well known, but other times difficult to detect, either directly or with standard statistical methods. Characteristics of well-known behavior patterns were abstracted and combined in order to define a scale-independent, hierarchical time pattern type, called a T-pattern. A corresponding detection algorithm was developed and implemented in a computer program, called Theme. The proposed pattern typology and detection algorithm are based on the definition and detection of a particular relationship between pairs of events in a time series, called a critical interval relation. The proposed bottom-up, level-by-level (or breadth-first) search algorithm is based on a binary tree of such relations. The algorithm first detects simpler patterns. Then, more complex and complete patterns evolve through the connection of simpler ones, pattern completeness competition, and pattern selection. Interindividual T-patterns in a quarter-hour interaction between two children are presented, showing that complex hidden T-patterns may be found by Theme in such behavioral streams. Finally, implications for studies of complexity, self-organization, and dynamic patterns are discussed.

The importance of temporal structure in analyzing schizophrenic behavior: some theoretical and diagnostic implications.

THEME, a new method for analyzing the temporal structure of responding on a two-choice task, is described. This method reveals the time relationships (temporal patterns) between all response events, even those not occurring in direct sequence. It selects those temporal patterns that are significantly different (p < 0.0001) from the patterns found in a random Poisson distribution of the same events. The method was applied to data from Lyon et al. (1986) in which n = 17 outpatient schizophrenics were compared with n = 17 age-, sex-, and education-matched normal control subjects. Results revealed that responding of schizophrenic outpatients, in comparison to control subjects, had a larger number of significant temporal patterns, more different types of patterns, and more branching (connectivity) of patterns at a higher level. The latter indicates a higher degree of internal structure. These results are not predicted by standard (DSM-III-R) diagnostic procedures, but are in agreement with studies of two-choice behavior in schizophrenia based on the Lyon-Robbins (1975) theory of behavioral change, which has possible relationship to dopamine/acetylcholine imbalance in the brain. Diagnostic procedures in schizophrenia might benefit from tests oriented toward these findings, which are also consistent with Bleuler's original descriptions of schizophrenic symptomatology.

Male-female interactions in staggerer and non-mutant mice: impairment to react to novelty as a possible explanation of staggerer male social behaviour.

Associated with neurological anomalies, many behavioural deficits are induced by the staggerer mutation. In order to define the consequences of this mutation on the staggerer male social behaviour we realised experimental dyadic encounters with non-mutant and unfamiliar females, either in estrous or in anestrous condition. We compared mutant behaviour to non-mutant male behaviour. Staggerer male behaviour presents the same characteristics during encounters with both types of females. It differs from non-mutant male behaviour in a similar context. Non-mutant males present more interactions (social interest, sniffing and sexual behaviours) with females than staggerer males. Females modify their behaviour as a function of their state of receptivity and according to their partner. Behaviours of staggerer males towards females may be interpreted in terms of general exploratory behaviour.

Effect of post-weaning social experience with normal females on the behaviour of adult male staggerer mice interacting with normal females.

Adult male staggerer mice reared under standard conditions display no sexual behaviour. When maintained for a time with normal female mice however 5% of the mutant males were able to copulate. We hypothesized that the effects of such social experience would be revealed most clearly by changes in the pattern of male behaviour. Two groups of mutant males were subjected to different social experience after weaning. One group was maintained with normal female mice and the other with staggerer mutant females. There were differences between the two groups in the frequency, duration and temporal organization of interactions between mutant males and normal females. Social experience with normal females modified male behaviour, producing greater synchrony with female behaviour and a reduction in stereotypic behaviours such as scratching.