Quantifying animal social behavior during pair bond formation 公开

Abstract

Pair bonding is a complex social process that involves behavioral changes across contexts, including sociosexual interactions and parental care. A pair bond is thus a complex state that manifests across an animal’s behavioral repertoire. Despite such complexity, this state is typically characterized with reference to only a few behaviors, often only one at a time. For example, in prairie voles, a canonical mammalian model of pair bonding, a bond is usually inferred only by observing one behavior, partner huddling, or another resident-intruder attack. Here, we characterize how the process of pair bonding alters a complete repertoire of natural behaviors. Specifically, we investigate prairie voles with their oxytocin receptors knocked out since the receptors are thought to be necessary for pair bond formation.

We tracked the postures of voles during the early phase (3hrs) of a 48-hr cohabitation by leveraging the advances in deep-learning algorithms such as DeepLabCut. However, these algorithms exhibit poor tracking accuracy when the animals spend more time with each other, performing key social behaviors such as huddling and mutual grooming. Hence, we implemented a pipeline that combines multiple deep-learning-based tracking methods to obtain detailed and high-accuracy postural trajectories of multiple animals.

Using the postural trajectories, we implemented an unsupervised method to map the behavior of the animals and observed that prairie voles show a gradual temporal evolution of the behavioral repertoire from locomotory-like movements to more stationary-like behaviors. In addition, we found that the mutant voles, with the oxytocin receptors knocked out, show more stationary-like behavior than wild-type animals. However, contrary to the current knowledge in the field and despite the subtle behavioral difference we observed, we found that the oxytocin receptors are not necessary for pair bond formation in prairie voles.

Table of Contents

LIST OF FIGURES ............................................................................................................. VI 

LIST OF SYMBOLS AND ABBREVIATIONS ................................................................... XIII 

Chapter 1. UNDERSTANDING THE ROLE OF THE OXYTOCIN SYSTEM IN PAIR BONDING THROUGH THE LENS OF THE PRAIRIE VOLE MODEL .............................. 1 

1.1 Introduction ......................................................................................................................1 

1.2 Background ......................................................................................................................1 

1.3 Social Monogamy..............................................................................................................2 

1.4 Prairie Voles: A Premier Animal Model for Pair Bonding ................................................4 

1.5 Oxytocin ...........................................................................................................................6 

1.6 Oxytocin Receptors ...........................................................................................................9 

1.7 Influence of the Oxytocin System in the Social Brain Network ....................................... 12 

1.8 Arginine Vasopressin ...................................................................................................... 16 

1.9 Dopamine ....................................................................................................................... 17 

1.10 Opioids ........................................................................................................................... 18 

1.11 Endocannabinoids .......................................................................................................... 19 

1.12 The social salience hypothesis ......................................................................................... 19 

1.12.1 Social Sensory Stimulus Transmission ........................................................................ 21 

1.12.2 Social Recognition and Memory ................................................................................. 22 

1.12.3 Social Gating .............................................................................................................. 23 

1.13 Concluding Remarks ...................................................................................................... 24 

Chapter 2. THE ENIGMA OF QUANTIFYING SOCIAL BEHAVIOR ............................... 25 

2.1 Introduction .................................................................................................................... 25 

2.2 Background .................................................................................................................... 25 

2.3 The question of social behavior ....................................................................................... 29 

2.4 Social Behavior ............................................................................................................... 30 

2.5 Framework for measuring social behavior...................................................................... 31 

2.6 Developing a pipeline for the framework ........................................................................ 32 

2.6.1 Videography ............................................................................................................... 33 

2.6.2 Tracking ..................................................................................................................... 34 

2.7 Extracting ethograms ..................................................................................................... 35 

2.8 Mapping behavior into a space ....................................................................................... 37 

2.9 Measuring hierarchy and predictability in social behaviors ............................................ 42 

2.10 Why care about measuring social behavior? ................................................................... 44 

2.11 Concluding Remarks ...................................................................................................... 45 

Chapter 3. DEVELOPING A PIPELINE FOR ROBUSTLY AND PRECISELY TRACKING MULTIPLE ANIMALS IN A SOCIAL CONTEXT .............................................................. 46 

3.1 Introduction .................................................................................................................... 46 

3.2 Background .................................................................................................................... 46 

3.3 Behavior tracking using computer vision and machine learning ..................................... 49 

3.4 Behavior tracking with deep learning approaches .......................................................... 51 

3.5 Pipeline to fix pose estimation ......................................................................................... 56 

3.5.1 Autoencoders .............................................................................................................. 57 

3.5.2 Autoencoder Architecture .......................................................................................... 58 

3.5.3 Type of Autoencoders ................................................................................................. 60 

3.5.4 The structure of the data input ................................................................................... 61 

3.5.4 The number of layers .................................................................................................. 61 

3.5.6 The number of neurons in each layer .......................................................................... 62 

3.5.7 The activation functions .............................................................................................. 63 

3.5.8 Regularizing the network ............................................................................................ 63 

3.5.8.1 Batch Normalization ................................................................................................... 64 

3.5.8.2 Dropout ...................................................................................................................... 64 

3.5.9 Loss function .............................................................................................................. 66 

3.5.10 Optimizers .................................................................................................................. 66 

3.6 Results and Discussion .................................................................................................... 67 

3.11 Conclusion ...................................................................................................................... 73 

Chapter 4. QUANTIFYING THE REPERTOIRE OF BEHAVIORS INVOLVED IN PRAIRIE VOLE SOCIAL BOND FORMATION ................................................................ 76 

4.1 Introduction .................................................................................................................... 76 

4.2 Background .................................................................................................................... 76 

4.3 Materials and Methods ................................................................................................... 81 

4.3.1 Experimental Model and Subject Details .................................................................... 81 

4.3.2 CRISPR/Cas9 knockout prairie voles ......................................................................... 81 

4.3.3 Animal Types .............................................................................................................. 82 

4.3.4 Behavioral Paradigm .................................................................................................. 82 

4.3.5 Tracking animals ........................................................................................................ 84 

4.3.6 Unsupervised mapping of behaviors ........................................................................... 86 

4.3.6.1 Transforming input data for MotionMapper .............................................................. 87 

4.3.6.2 Overview of modified MotionMapper pipeline ........................................................... 89 

4.3.7 Supervised extraction of behavior ............................................................................... 90 

4.3.8 Statistical analysis ....................................................................................................... 91 

4.4 Results ............................................................................................................................ 91 

4.4.1 Oxytocin Receptor Knockout voles can form social bonds .......................................... 91 

.................................................................................................................................................... 93 

4.4.2 Knocking out OTRs does not impair social curiosity and locomotory behavior .......... 93 

4.4.3 Examining the behavioral repertoire of prairie voles during the early phase of cohabitation ................................................................................................................................ 96 

4.4.4 CRISPR KO animals show subtle behavioral differences as compared to WT animals 100 

4.4.5 Subtle differences between CRISPR KO and WT animals show up in the early stages of bonding ................................................................................................................................. 101 

4.4.6 The subtle behavioral difference between CRISPR KO and WT associated with oral investigation and movement activities ....................................................................................... 105 

4.5 DISCUSSION ............................................................................................................... 108 

Chapter 5. CONCLUSION AND FUTURE DIRECTIONS ................................................ 112 

5.1 Thesis Contribution ...................................................................................................... 112 

5.2 Summaries of Chapters ................................................................................................ 113 

5.2.1 Chapter 1 summary .................................................................................................. 113 

5.2.2 Chapter 2 summary .................................................................................................. 114 

5.2.3 Chapter 3 summary .................................................................................................. 114 

5.2.3 Chapter 4 summary .................................................................................................. 115 

5.3 Future Directions .......................................................................................................... 116 

5.3.1 Beyond Videography ................................................................................................ 116 

5.3.2 What about non-stereotyped behaviors? ................................................................... 117 

5.3.3 Extending from 2D to a 3D recording of social behavior ........................................... 117 

5.3.4 Recording electrophysiological activity during social interaction .............................. 118 

5.3.5 How about the arginine vasopressin system? ............................................................ 119 

5.3.6 How much deeper can we go? ................................................................................... 121 

Chapter 6. APPENDIX A. CHAPTER 4 SUPPLEMENTARY MATERIAL ...................................... 122 

A.1 Meta Data of prairie voles used in experiments............................................................. 122 

A.2 Supplemental Figures ................................................................................................... 135 

Reference .......................................................................................................................... 139 

About this Dissertation

Rights statement

• Permission granted by the author to include this thesis or dissertation in this repository. All rights reserved by the author. Please contact the author for information regarding the reproduction and use of this thesis or dissertation.

School	Laney Graduate School
Department	Biological and Biomedical Sciences
Subfield / Discipline	Neuroscience
Degree	Ph.D.
Submission	Dissertation
Language	English
Research Field	Biology, Neuroscience
关键词	Oxytocin Pair bonding Computational Neuroscience Deep Learning Machine Learning Social Behavior Prairie voles
Committee Chair / Thesis Advisor	Berman, Gordon, Emory University Liu, Robert, Emory University
Committee Members	Young, Larry, Emory University Kelly, Aubrey, Emory University Sober, Samuel, Emory University

最新修改

Primary PDF

Thumbnail	Title	Date Uploaded	Actions
	Quantifying animal social behavior during pair bond formation ()	2023-03-07 14:55:38 -0500	Press to Select an action Download

Supplemental Files

Thumbnail	Title	Date Uploaded	Actions