diff --git a/VRC_PonderRNN.ipynb b/VRC_PonderRNN.ipynb
index 8dc13b0..98a30aa 100644
--- a/VRC_PonderRNN.ipynb
+++ b/VRC_PonderRNN.ipynb
@@ -3,19 +3,23 @@
   {
    "cell_type": "markdown",
    "source": [
-    "Building on [PonderNet](https://arxiv.org/abs/2107.05407), this notebook implements a neural alternative of the  [Variable Rate Coding](https://doi.org/10.32470/CCN.2019.1397-0) model to produce human-like responses.\n",
+    "## Intro\n",
+    "\n",
+    "In the context of behavioral data, we are interested in simultaneously modeling speed and accuracy. Yet, most advanced techniques in machine learning cannot capture such a duality of decision making data.\n",
+    "\n",
+    "\n",
+    "Building on [PonderNet](https://arxiv.org/abs/2107.05407) and [Variable Rate Coding](https://doi.org/10.32470/CCN.2019.1397-0), this notebook implements a neural model that captures speed and accuracy of human-like responses.\n",
     "\n",
     "Given stimulus symbols as inputs, the model produces two outputs:\n",
     "\n",
     "- Response symbol, which, in comparison with the input stimuli, can be used to measure accuracy).\n",
-    "- Remaining entropy (to be contrasted against a decision threshold and ultimateely halt the process).\n",
+    "- Halting probability ($\\lambda_n$).\n",
     "\n",
-    "Under the hood, the model uses a RNN along with multiple Poisson processes to...\n",
+    "Under the hood, the model iterates over a ICOM-like component to reach a halting point in time. Unlike DDM and ICOM models, all the parameters and outcomes of the current model *seem* cognitively interpretable.\n",
     "\n",
+    "### Additional resources\n",
     "\n",
-    "## Resources\n",
-    "\n",
-    "- [Network model](https://drive.google.com/file/d/16eiUUwKGWfh9pu9VUxzlx046hQNHV0Qe/view?usp=sharinghttps://drive.google.com/file/d/16eiUUwKGWfh9pu9VUxzlx046hQNHV0Qe/view?usp=sharing)\n"
+    "- [ICOM network model](https://drive.google.com/file/d/16eiUUwKGWfh9pu9VUxzlx046hQNHV0Qe/view?usp=sharinghttps://drive.google.com/file/d/16eiUUwKGWfh9pu9VUxzlx046hQNHV0Qe/view?usp=sharing)\n"
    ],
    "metadata": {}
   },
@@ -369,9 +373,33 @@
   },
   {
    "cell_type": "code",
-   "execution_count": null,
-   "source": [],
-   "outputs": [],
+   "execution_count": 45,
+   "source": [
+    "# example code to decode a stimulus into multiple sequence (one per channel)\n",
+    "\n",
+    "import torch\n",
+    "from torch import nn\n",
+    "\n",
+    "n_inputs = 7\n",
+    "max_timestep = 10\n",
+    "n_channels = 5\n",
+    "\n",
+    "X = torch.nn.functional.one_hot(torch.tensor(4), num_classes=n_inputs).type(torch.float)\n",
+    "\n",
+    "decode = nn.Linear(n_inputs, n_channels * max_timestep)\n",
+    "out = decode(X).reshape((n_channels, max_timestep))\n",
+    "\n",
+    "print(out.shape)"
+   ],
+   "outputs": [
+    {
+     "output_type": "stream",
+     "name": "stdout",
+     "text": [
+      "torch.Size([5, 10])\n"
+     ]
+    }
+   ],
    "metadata": {}
   }
  ],