Introduction

The eShepherd system is built on decades of scientific research into how cattle learn and respond to their environment.

This research has resulted in a patented training algorithm that is designed to provide an animal with predictability and control. Whilst this automated training program reduces the need for operator intervention, it does require an understanding of how cattle think, how they learn, and what signs tell you that the system is working as intended or that an animal needs attention.

This guide covers:

• How cattle learn to respond to the virtual fence
• What good training looks like
• Animal welfare indicators to monitor
• What to do when something goes wrong

Core principle: The eShepherd system works with natural animal learning processes, not against them. When it is set up and operated correctly, cattle experience low and predictable levels of aversive stimulus. Your role as an operator is to create conditions where animals can learn quickly, comfortably, and reliably.

How Cattle Learn — The Science Behind eShepherd

eShepherd uses two well-established principles of animal learning.

Classical conditioning is learning through association. A previously neutral stimulus (the audio cue) is repeatedly paired with an aversive stimulus (the pulse) until the animal responds to the audio cue alone. This is the same mechanism by which cattle learn to avoid an electric fence after one or two contacts.

Operant conditioning is learning through consequences. The animal discovers that a specific behaviour (turning away from the boundary) removes the unpleasant consequence (the pulse). The animal gains control over its own experience. This is critical to animal welfare — an animal that understands how to avoid discomfort is not a stressed animal.

Together, these two processes produce a conditioned animal that responds to the audio cue alone and rarely requires a pulse stimulus at all.

The Cognitive Activation Theory of Stress (CATS)

The eShepherd training algorithm is designed around the CATS framework (Lee, Colditz & Campbell, 2018). CATS describes how an animal's stress response depends on its ability to predict and control what happens to it.

An animal that:

• Can predict when a stimulus will occur (high predictability), and
• Can control whether it receives that stimulus (high control)

…is a low-stress animal.

An animal that receives random, unpredictable, or unavoidable stimuli is a high-stress animal — and is also one that will not learn effectively.

Practical implication: Everything about how you set up and manage your virtual paddocks affects the animal's ability to predict and avoid the boundary. Poor paddock design or rushed training creates unpredictability and reduces control. This impacts both animal welfare and system effectiveness.

What Happens During Training

The training period lasts between 7–10 days for a mob of up to 500 head, but note that the animals' learning does not stop there. They will continue to become more skilled at navigating virtual fence boundaries as time passes. In every mob there are 'leaders' and followers'. The 'followers' tend to stay away from the virtual boundary and observe the responses from the 'leaders'. Hence, it may take longer for those 'followers' to learn how to respond correctly to avoid the pulse stimulus.

During the training period:

1. Animals are introduced to a simple Virtual Paddock with a simple fence segments (no complex boundaries).
2. As animals approach the boundary, they hear the audio cue.
3. Animals that do not respond receive the aversive pulse stimulus.
4. Over time, animals learn to turn away at the audio cue alone.
5. The eShepherd system monitors each animal's learning metrics automatically (called the Audio Ratio described below)
6. When training metrics are met, the platform will display this information, confirming the mob is ready for a more complex Virtual Paddock.

Setting Up for Successful Training

Before activating the training paddock, confirm the following:

• All animals are fitted with a Neckband
• All Neckbands and animals are registered in the eShepherd Web App
• The training paddock design has been discussed with the eShepherd CS Team
• The training paddock is free from dense shrub
• Physical boundary fences in the training paddock are in good repair
• There is sufficient feed and a working water point for 7–10 days
• Animals have acclimatised to the paddock and are grazing or resting calmly
• A simple Virtual Paddock has been created for the training paddock

Designing the Training Virtual Paddock

Good paddock design is one of the most important welfare controls in the system.

Rules for the training Virtual Paddock:

• Draw the paddock so animals are likely to interact with only one fence segment at a time
• Where a segment has corners, keep angles greater than 145 degrees (closer to 180 is better)
• Set the fence at least 50 m from any water point, feed supplement, or shelter
• Where the Virtual Paddock runs near a physical fence, set it at least 50 m inside or outside the physical fence
• Allow animals to move 20–100 m into the Exclusion Zone — this gives them room to learn and turn around without becoming trapped

Why this matters: Sharp corners, proximity to water, or inadequate space in the Exclusion Zone create situations where animals receive confusing or unavoidable cues. This increases pulse count, increases stress, and slows learning.

Updated on: 17/03/2026