Nback

[jdcpni]

• Project

• get rid of keywords for MSE, SSE, L0, L1, etc. and rename as Loss.<*>.name
• NOTE: keyword for CROSS_ENTROPY ('cross-entropy') persists,
  as it is used for DistanceMetrics and LinearCombination.operation

• composition.py, learningfunctions.py

• loss_function -> loss_spec (for consistency with autodiff)

• outputport.py

• StandardOutputPorts: fix bug in get_port_dict in which name searched for with "is" rather than "=="

• comparatormechanism.py

• rename standard_output_port['sum'] as standard_output_port['SUM']
• NOTE: names of MSE and SSE output_ports use Loss.<*>.name, where as SUM uses SUM.upper() (instead of L0)

[github-actions]

This PR causes the following changes to the html docs (ubuntu-latest-3.7-x64):

diff -r docs-base/AutodiffComposition.html docs-head/AutodiffComposition.html
293,294c293,294
< This <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a> provides a summary and comparison of these different modes of execution,
< that are described in greater detail below.</p>
---
> These are each described in greater detail below, and summarized in this <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a>
> which provides a comparison of the different modes of execution for an AutodiffComposition and standard <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a>.</p>
325c325
< executing using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
---
> executed using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
327a328,341
> <div class="admonition warning">
> <p class="admonition-title">Warning</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.LLVM" title="psyneulink.core.llvm.__init__.ExecutionMode.LLVM"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.LLVM</span></code></a> or <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> in the learn() method of a standard
> <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a> causes an error.</p></li>
> </ul>
> </div>
> <div class="admonition note">
> <p class="admonition-title">Note</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.Python" title="psyneulink.core.llvm.__init__.ExecutionMode.Python"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.Python</span></code></a> in the learn() method of a <a class="reference internal" href="#"><span class="doc">AutodiffComposition</span></a> is treated as a
> synonym of <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> (see table).</p></li>
> </ul>
> </div>
415a430,452
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
> <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
> <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
> arguments from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch optimizer function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
> <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
> <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch loss function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
418c455
< <dd><p>tracks the average for each weight update (i.e. each minibatch)</p>
---
> <dd><p>tracks the average loss after each weight update (i.e. each minibatch) during learning.</p>
427,430c464,466
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
< <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
< <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
< arguments from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights">
> <span class="sig-name descname"><span class="pre">last_saved_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file to which weights were last saved.</p>
433c469
< <dd class="field-odd"><p>PyTorch optimizer function</p>
---
> <dd class="field-odd"><p>path</p>
439,441c475,477
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
< <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
< <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights">
> <span class="sig-name descname"><span class="pre">last_loaded_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file from which weights were last loaded.</p>
444c480
< <dd class="field-odd"><p>PyTorch loss function</p>
---
> <dd class="field-odd"><p>path</p>
457c493
< <dd><p>Updates parameters based on trials run since last update.</p>
---
> <dd><p>Updates parameters (weights) based on trials run since last update.</p>
636c672
< <dd><p>Loads all weights matrices for all MappingProjections in the AutodiffComposition from file
---
> <dd><p>Loads all weight matrices for all MappingProjections in the AutodiffComposition from file
diff -r docs-base/BasicsAndPrimer.html docs-head/BasicsAndPrimer.html
259c259
< <span id="basicsandprimer-grandview-figure"></span><img alt="_static/BasicsAndPrimer_GrandView_fig.svg" src="_static/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
---
> <span id="basicsandprimer-grandview-figure"></span><img alt="_images/BasicsAndPrimer_GrandView_fig.svg" src="_images/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
269,270c269,270
< <code class="xref any docutils literal notranslate"><span class="pre">tutorial</span></code> provides additional introductory material for those who are newer to computational modeling, as well as a
< more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
---
> <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">tutorial</span></a> provides additional introductory material for those who are newer to computational modeling,
> as well as a more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
diff -r docs-base/BeukersNBackModel.html docs-head/BeukersNBackModel.html
13c13
<   <title>N-Back Model (Beukers et al., 2022) &mdash; PsyNeuLink 0.0.0.0 documentation</title>
---
>   <title>Nback Model &mdash; PsyNeuLink 0.0.0.0 documentation</title>
35c35,37
<     <link rel="search" title="Search" href="search.html" /> 
---
>     <link rel="search" title="Search" href="search.html" />
>     <link rel="next" title="Contributors Guide" href="ContributorsGuide.html" />
>     <link rel="prev" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" href="BustamanteStroopXORLVOCModel.html" /> 
129c131
<               <ul>
---
>               <ul class="current">
134c136
< <li class="toctree-l1"><a class="reference internal" href="Library.html">Library</a></li>
---
> <li class="toctree-l1 current"><a class="reference internal" href="Library.html">Library</a></li>
177c179,183
<       <li>N-Back Model (Beukers et al., 2022)</li>
---
>           <li><a href="Library.html">Library</a> &gt;</li>
>         
>           <li><a href="Models.html">Models</a> &gt;</li>
>         
>       <li>Nback Model</li>
200,268c206,213
<   <section id="n-back-model-beukers-et-al-2022">
< <h1>N-Back Model (Beukers et al., 2022)<a class="headerlink" href="#n-back-model-beukers-et-al-2022" title="Permalink to this headline">¶</a></h1>
< <p><a class="reference external" href="https://psyarxiv.com/jtw5p">“When Working Memory is Just Working, Not Memory”</a></p>
< <section id="overview">
< <h2>Overview<a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h2>
< <p>This implements a model of the <a class="reference external" href="https://en.wikipedia.org/wiki/N-back#Neurobiology_of_n-back_task">N-back task</a>
< described in <a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al. (2022)</a>.  The model uses a simple implementation of episodic
< memory (EM, as a form of content-retrieval memory) to store previous stimuli along with the temporal context in which
< they occured, and a feedforward neural network (FFN)to evaluate whether the current stimulus is a match to the n’th
< preceding stimulus (nback-level)retrieved from episodic memory.  The temporal context is provided by a randomly
< drifting high dimensional vector that maintains a constant norm (i.e., drifts on a sphere).  The FFN is
< trained, given an n-back level of <em>n</em>, to identify when the current stimulus matches one stored in EM
< with a temporal context vector that differs by an amount corresponding to <em>n</em> time steps of drift.  During n-back
< performance, the model encodes the current stimulus and temporal context, retrieves an item from EM that matches the
< current stimulus, weighted by the similarity of its temporal context vector (i.e., most recent), and then uses the
< FFN to evaluate whether it is an n-back match.  The model responds “match” if the FFN detects a match; otherwise, it
< either responds “non-match” or, with a fixed probability (hazard rate), it uses the current stimulus and temporal
< context to retrieve another sample from EM and repeat the evaluation.</p>
< <p>This model is an example of proposed interactions between working memory (e.g., in neocortex) and episodic memory
< e.g., in hippocampus and/or cerebellum) in the performance of tasks demanding of sequential processing and control,
< and along the lines of models emerging machine learning that augment the use of recurrent neural networks (e.g., long
< short-term memory mechanisms; LSTMs) for active memory and control with an external memory capable of rapid storage
< and content-based retrieval, such as the Neural Turing Machine (NTN;
< <a class="reference external" href="https://arxiv.org/abs/1410.5401">Graves et al., 2016</a>), Episodic Planning Networks (EPN;
< <a class="reference external" href="https://arxiv.org/abs/2006.03662">Ritter et al., 2020</a>), and Emergent Symbols through Binding Networks (ESBN;
< <a class="reference external" href="https://arxiv.org/abs/2012.14601">Webb et al., 2021</a>).</p>
< <p>The script respectively, to construct, train and run the model:</p>
< <ul class="simple">
< <li><p>construct_model(args):
< takes as arguments parameters used to construct the model; for convenience, defaults are defined toward the top
< of the script (see “Construction parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>train_network(args)
< takes as arguments the feedforward neural network Composition (FFN_COMPOSITION) and number of epochs to train.
< Note: learning_rate is set at construction (which can be specified using LEARNING_RATE under “Training parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>run_model()
< takes as arguments the drift rate in the temporal context vector to be applied on each trial,
< and the number of trials to execute, as well as reporting and animation specifications
< (see “Execution parameters”).</p></li>
< </ul>
< <p>The default parameters are ones that have been fit to empirical data concerning human performance
< (taken from <a class="reference external" href="https://psycnet.apa.org/record/2007-06096-010?doi=1">Kane et al., 2007</a>).</p>
< </section>
< <section id="the-model">
< <h2>The Model<a class="headerlink" href="#the-model" title="Permalink to this headline">¶</a></h2>
< <p>The models is composed of two <a class="reference internal" href="Composition.html"><span class="doc">Compositions</span></a>: an outer one that contains the full model (nback_model),
< and an <a class="reference internal" href="AutodiffComposition.html"><span class="doc">AutodiffComposition</span></a> (ffn), nested within nback_model (see red box in Figure), that implements the
< feedforward neural network (ffn).</p>
< <section id="nback-model">
< <h3>nback_model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h3>
< <p>This contains three input Mechanisms (</p>
< <p>Both of these are constructed in the construct_model function.
< The ffn Composition is trained use</p>
< <figure class="align-left" id="nback-fig">
< <img alt="N-Back Model Animation" src="_images/N-Back_Model_movie.gif" />
< </figure>
< </section>
< </section>
< <section id="training">
< <h2>Training<a class="headerlink" href="#training" title="Permalink to this headline">¶</a></h2>
< </section>
< <section id="execution">
< <h2>Execution<a class="headerlink" href="#execution" title="Permalink to this headline">¶</a></h2>
< <p>Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code>
< .. Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code></p>
< </section>
---
>   <section id="nback-model">
> <h1>Nback Model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h1>
> <p><strong>“When Working Memory is Just Working, Not Memory”</strong> (<a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al., 2022</a>)</p>
> <p>Script: <code class="xref download docutils literal notranslate"><span class="pre">nback.py</span></code></p>
> <p>Example of Jupyter Notebook: <a class="reference internal" href="BeukersNBackModel_NB.html"><span class="doc">Nback Model Notebook</span></a></p>
> <p>(Instructions for running the actual notebook can be found <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">here</span></a>, replacing <code class="docutils literal notranslate"><span class="pre">nback_nb</span></code> for <code class="docutils literal notranslate"><span class="pre">tutorial</span></code>)</p>
> <div class="toctree-wrapper compound">
> </div>
276a222,230
>     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
>       
>         <a href="ContributorsGuide.html" class="btn btn-neutral float-right" title="Contributors Guide" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
>       
>       
>         <a href="BustamanteStroopXORLVOCModel.html" class="btn btn-neutral" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
>       
>     </div>
>   
305,314c259
< <li><a class="reference internal" href="#">N-Back Model (Beukers et al., 2022)</a><ul>
< <li><a class="reference internal" href="#overview">Overview</a></li>
< <li><a cla
...

See CI logs for the full diff.

[github-advanced-security]

CodeQL found more than 10 potential problems in the proposed changes. Check the Files changed tab for more details.

[github-actions]

This PR causes the following changes to the html docs (ubuntu-latest-3.7-x64):

diff -r docs-base/AutodiffComposition.html docs-head/AutodiffComposition.html
293,294c293,294
< This <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a> provides a summary and comparison of these different modes of execution,
< that are described in greater detail below.</p>
---
> These are each described in greater detail below, and summarized in this <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a>
> which provides a comparison of the different modes of execution for an AutodiffComposition and standard <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a>.</p>
325c325
< executing using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
---
> executed using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
327a328,341
> <div class="admonition warning">
> <p class="admonition-title">Warning</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.LLVM" title="psyneulink.core.llvm.__init__.ExecutionMode.LLVM"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.LLVM</span></code></a> or <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> in the learn() method of a standard
> <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a> causes an error.</p></li>
> </ul>
> </div>
> <div class="admonition note">
> <p class="admonition-title">Note</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.Python" title="psyneulink.core.llvm.__init__.ExecutionMode.Python"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.Python</span></code></a> in the learn() method of a <a class="reference internal" href="#"><span class="doc">AutodiffComposition</span></a> is treated as a
> synonym of <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> (see table).</p></li>
> </ul>
> </div>
415a430,452
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
> <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
> <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
> arguments from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch optimizer function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
> <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
> <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch loss function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
418c455
< <dd><p>tracks the average for each weight update (i.e. each minibatch)</p>
---
> <dd><p>tracks the average loss after each weight update (i.e. each minibatch) during learning.</p>
427,430c464,466
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
< <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
< <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
< arguments from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights">
> <span class="sig-name descname"><span class="pre">last_saved_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file to which weights were last saved.</p>
433c469
< <dd class="field-odd"><p>PyTorch optimizer function</p>
---
> <dd class="field-odd"><p>path</p>
439,441c475,477
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
< <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
< <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights">
> <span class="sig-name descname"><span class="pre">last_loaded_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file from which weights were last loaded.</p>
444c480
< <dd class="field-odd"><p>PyTorch loss function</p>
---
> <dd class="field-odd"><p>path</p>
457c493
< <dd><p>Updates parameters based on trials run since last update.</p>
---
> <dd><p>Updates parameters (weights) based on trials run since last update.</p>
636c672
< <dd><p>Loads all weights matrices for all MappingProjections in the AutodiffComposition from file
---
> <dd><p>Loads all weight matrices for all MappingProjections in the AutodiffComposition from file
diff -r docs-base/BasicsAndPrimer.html docs-head/BasicsAndPrimer.html
259c259
< <span id="basicsandprimer-grandview-figure"></span><img alt="_static/BasicsAndPrimer_GrandView_fig.svg" src="_static/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
---
> <span id="basicsandprimer-grandview-figure"></span><img alt="_images/BasicsAndPrimer_GrandView_fig.svg" src="_images/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
269,270c269,270
< <code class="xref any docutils literal notranslate"><span class="pre">tutorial</span></code> provides additional introductory material for those who are newer to computational modeling, as well as a
< more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
---
> <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">tutorial</span></a> provides additional introductory material for those who are newer to computational modeling,
> as well as a more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
diff -r docs-base/BeukersNBackModel.html docs-head/BeukersNBackModel.html
13c13
<   <title>N-Back Model (Beukers et al., 2022) &mdash; PsyNeuLink 0.0.0.0 documentation</title>
---
>   <title>Nback Model &mdash; PsyNeuLink 0.0.0.0 documentation</title>
35c35,37
<     <link rel="search" title="Search" href="search.html" /> 
---
>     <link rel="search" title="Search" href="search.html" />
>     <link rel="next" title="Contributors Guide" href="ContributorsGuide.html" />
>     <link rel="prev" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" href="BustamanteStroopXORLVOCModel.html" /> 
129c131
<               <ul>
---
>               <ul class="current">
134c136
< <li class="toctree-l1"><a class="reference internal" href="Library.html">Library</a></li>
---
> <li class="toctree-l1 current"><a class="reference internal" href="Library.html">Library</a></li>
177c179,183
<       <li>N-Back Model (Beukers et al., 2022)</li>
---
>           <li><a href="Library.html">Library</a> &gt;</li>
>         
>           <li><a href="Models.html">Models</a> &gt;</li>
>         
>       <li>Nback Model</li>
200,268c206,213
<   <section id="n-back-model-beukers-et-al-2022">
< <h1>N-Back Model (Beukers et al., 2022)<a class="headerlink" href="#n-back-model-beukers-et-al-2022" title="Permalink to this headline">¶</a></h1>
< <p><a class="reference external" href="https://psyarxiv.com/jtw5p">“When Working Memory is Just Working, Not Memory”</a></p>
< <section id="overview">
< <h2>Overview<a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h2>
< <p>This implements a model of the <a class="reference external" href="https://en.wikipedia.org/wiki/N-back#Neurobiology_of_n-back_task">N-back task</a>
< described in <a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al. (2022)</a>.  The model uses a simple implementation of episodic
< memory (EM, as a form of content-retrieval memory) to store previous stimuli along with the temporal context in which
< they occured, and a feedforward neural network (FFN)to evaluate whether the current stimulus is a match to the n’th
< preceding stimulus (nback-level)retrieved from episodic memory.  The temporal context is provided by a randomly
< drifting high dimensional vector that maintains a constant norm (i.e., drifts on a sphere).  The FFN is
< trained, given an n-back level of <em>n</em>, to identify when the current stimulus matches one stored in EM
< with a temporal context vector that differs by an amount corresponding to <em>n</em> time steps of drift.  During n-back
< performance, the model encodes the current stimulus and temporal context, retrieves an item from EM that matches the
< current stimulus, weighted by the similarity of its temporal context vector (i.e., most recent), and then uses the
< FFN to evaluate whether it is an n-back match.  The model responds “match” if the FFN detects a match; otherwise, it
< either responds “non-match” or, with a fixed probability (hazard rate), it uses the current stimulus and temporal
< context to retrieve another sample from EM and repeat the evaluation.</p>
< <p>This model is an example of proposed interactions between working memory (e.g., in neocortex) and episodic memory
< e.g., in hippocampus and/or cerebellum) in the performance of tasks demanding of sequential processing and control,
< and along the lines of models emerging machine learning that augment the use of recurrent neural networks (e.g., long
< short-term memory mechanisms; LSTMs) for active memory and control with an external memory capable of rapid storage
< and content-based retrieval, such as the Neural Turing Machine (NTN;
< <a class="reference external" href="https://arxiv.org/abs/1410.5401">Graves et al., 2016</a>), Episodic Planning Networks (EPN;
< <a class="reference external" href="https://arxiv.org/abs/2006.03662">Ritter et al., 2020</a>), and Emergent Symbols through Binding Networks (ESBN;
< <a class="reference external" href="https://arxiv.org/abs/2012.14601">Webb et al., 2021</a>).</p>
< <p>The script respectively, to construct, train and run the model:</p>
< <ul class="simple">
< <li><p>construct_model(args):
< takes as arguments parameters used to construct the model; for convenience, defaults are defined toward the top
< of the script (see “Construction parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>train_network(args)
< takes as arguments the feedforward neural network Composition (FFN_COMPOSITION) and number of epochs to train.
< Note: learning_rate is set at construction (which can be specified using LEARNING_RATE under “Training parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>run_model()
< takes as arguments the drift rate in the temporal context vector to be applied on each trial,
< and the number of trials to execute, as well as reporting and animation specifications
< (see “Execution parameters”).</p></li>
< </ul>
< <p>The default parameters are ones that have been fit to empirical data concerning human performance
< (taken from <a class="reference external" href="https://psycnet.apa.org/record/2007-06096-010?doi=1">Kane et al., 2007</a>).</p>
< </section>
< <section id="the-model">
< <h2>The Model<a class="headerlink" href="#the-model" title="Permalink to this headline">¶</a></h2>
< <p>The models is composed of two <a class="reference internal" href="Composition.html"><span class="doc">Compositions</span></a>: an outer one that contains the full model (nback_model),
< and an <a class="reference internal" href="AutodiffComposition.html"><span class="doc">AutodiffComposition</span></a> (ffn), nested within nback_model (see red box in Figure), that implements the
< feedforward neural network (ffn).</p>
< <section id="nback-model">
< <h3>nback_model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h3>
< <p>This contains three input Mechanisms (</p>
< <p>Both of these are constructed in the construct_model function.
< The ffn Composition is trained use</p>
< <figure class="align-left" id="nback-fig">
< <img alt="N-Back Model Animation" src="_images/N-Back_Model_movie.gif" />
< </figure>
< </section>
< </section>
< <section id="training">
< <h2>Training<a class="headerlink" href="#training" title="Permalink to this headline">¶</a></h2>
< </section>
< <section id="execution">
< <h2>Execution<a class="headerlink" href="#execution" title="Permalink to this headline">¶</a></h2>
< <p>Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code>
< .. Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code></p>
< </section>
---
>   <section id="nback-model">
> <h1>Nback Model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h1>
> <p><strong>“When Working Memory is Just Working, Not Memory”</strong> (<a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al., 2022</a>)</p>
> <p>Script: <a class="reference download internal" download="" href="_downloads/3ca6d8ab14ce9c785a6b5c1b1d60a955/nback.py"><code class="xref download docutils literal notranslate"><span class="pre">nback.py</span></code></a></p>
> <p>Example of Jupyter Notebook: <a class="reference internal" href="BeukersNBackModel_NB.html"><span class="doc">Nback Model Notebook</span></a></p>
> <p>(Instructions for running the actual notebook can be found <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">here</span></a>, replacing <code class="docutils literal notranslate"><span class="pre">nback_nb</span></code> for <code class="docutils literal notranslate"><span class="pre">tutorial</span></code>)</p>
> <div class="toctree-wrapper compound">
> </div>
276a222,230
>     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
>       
>         <a href="ContributorsGuide.html" class="btn btn-neutral float-right" title="Contributors Guide" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
>       
>       
>         <a href="BustamanteStroopXORLVOCModel.html" class="btn btn-neutral" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
>       
>     </div>
>   
305,314c259
< <li><a class="reference internal" href="#">N-Back Model
...

See CI logs for the full diff.

[github-actions]

This PR causes the following changes to the html docs (ubuntu-latest-3.7-x64):

diff -r docs-base/AutodiffComposition.html docs-head/AutodiffComposition.html
293,294c293,294
< This <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a> provides a summary and comparison of these different modes of execution,
< that are described in greater detail below.</p>
---
> These are each described in greater detail below, and summarized in this <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a>
> which provides a comparison of the different modes of execution for an AutodiffComposition and standard <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a>.</p>
325c325
< executing using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
---
> executed using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
327a328,341
> <div class="admonition warning">
> <p class="admonition-title">Warning</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.LLVM" title="psyneulink.core.llvm.__init__.ExecutionMode.LLVM"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.LLVM</span></code></a> or <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> in the learn() method of a standard
> <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a> causes an error.</p></li>
> </ul>
> </div>
> <div class="admonition note">
> <p class="admonition-title">Note</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.Python" title="psyneulink.core.llvm.__init__.ExecutionMode.Python"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.Python</span></code></a> in the learn() method of a <a class="reference internal" href="#"><span class="doc">AutodiffComposition</span></a> is treated as a
> synonym of <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> (see table).</p></li>
> </ul>
> </div>
415a430,452
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
> <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
> <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
> arguments from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch optimizer function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
> <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
> <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch loss function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
418c455
< <dd><p>tracks the average for each weight update (i.e. each minibatch)</p>
---
> <dd><p>tracks the average loss after each weight update (i.e. each minibatch) during learning.</p>
427,430c464,466
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
< <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
< <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
< arguments from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights">
> <span class="sig-name descname"><span class="pre">last_saved_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file to which weights were last saved.</p>
433c469
< <dd class="field-odd"><p>PyTorch optimizer function</p>
---
> <dd class="field-odd"><p>path</p>
439,441c475,477
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
< <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
< <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights">
> <span class="sig-name descname"><span class="pre">last_loaded_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file from which weights were last loaded.</p>
444c480
< <dd class="field-odd"><p>PyTorch loss function</p>
---
> <dd class="field-odd"><p>path</p>
457c493
< <dd><p>Updates parameters based on trials run since last update.</p>
---
> <dd><p>Updates parameters (weights) based on trials run since last update.</p>
636c672
< <dd><p>Loads all weights matrices for all MappingProjections in the AutodiffComposition from file
---
> <dd><p>Loads all weight matrices for all MappingProjections in the AutodiffComposition from file
diff -r docs-base/BasicsAndPrimer.html docs-head/BasicsAndPrimer.html
259c259
< <span id="basicsandprimer-grandview-figure"></span><img alt="_static/BasicsAndPrimer_GrandView_fig.svg" src="_static/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
---
> <span id="basicsandprimer-grandview-figure"></span><img alt="_images/BasicsAndPrimer_GrandView_fig.svg" src="_images/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
269,270c269,270
< <code class="xref any docutils literal notranslate"><span class="pre">tutorial</span></code> provides additional introductory material for those who are newer to computational modeling, as well as a
< more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
---
> <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">tutorial</span></a> provides additional introductory material for those who are newer to computational modeling,
> as well as a more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
diff -r docs-base/BeukersNBackModel.html docs-head/BeukersNBackModel.html
13c13
<   <title>N-Back Model (Beukers et al., 2022) &mdash; PsyNeuLink 0.0.0.0 documentation</title>
---
>   <title>Nback Model &mdash; PsyNeuLink 0.0.0.0 documentation</title>
35c35,37
<     <link rel="search" title="Search" href="search.html" /> 
---
>     <link rel="search" title="Search" href="search.html" />
>     <link rel="next" title="Contributors Guide" href="ContributorsGuide.html" />
>     <link rel="prev" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" href="BustamanteStroopXORLVOCModel.html" /> 
129c131
<               <ul>
---
>               <ul class="current">
134c136
< <li class="toctree-l1"><a class="reference internal" href="Library.html">Library</a></li>
---
> <li class="toctree-l1 current"><a class="reference internal" href="Library.html">Library</a></li>
177c179,183
<       <li>N-Back Model (Beukers et al., 2022)</li>
---
>           <li><a href="Library.html">Library</a> &gt;</li>
>         
>           <li><a href="Models.html">Models</a> &gt;</li>
>         
>       <li>Nback Model</li>
200,268c206,213
<   <section id="n-back-model-beukers-et-al-2022">
< <h1>N-Back Model (Beukers et al., 2022)<a class="headerlink" href="#n-back-model-beukers-et-al-2022" title="Permalink to this headline">¶</a></h1>
< <p><a class="reference external" href="https://psyarxiv.com/jtw5p">“When Working Memory is Just Working, Not Memory”</a></p>
< <section id="overview">
< <h2>Overview<a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h2>
< <p>This implements a model of the <a class="reference external" href="https://en.wikipedia.org/wiki/N-back#Neurobiology_of_n-back_task">N-back task</a>
< described in <a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al. (2022)</a>.  The model uses a simple implementation of episodic
< memory (EM, as a form of content-retrieval memory) to store previous stimuli along with the temporal context in which
< they occured, and a feedforward neural network (FFN)to evaluate whether the current stimulus is a match to the n’th
< preceding stimulus (nback-level)retrieved from episodic memory.  The temporal context is provided by a randomly
< drifting high dimensional vector that maintains a constant norm (i.e., drifts on a sphere).  The FFN is
< trained, given an n-back level of <em>n</em>, to identify when the current stimulus matches one stored in EM
< with a temporal context vector that differs by an amount corresponding to <em>n</em> time steps of drift.  During n-back
< performance, the model encodes the current stimulus and temporal context, retrieves an item from EM that matches the
< current stimulus, weighted by the similarity of its temporal context vector (i.e., most recent), and then uses the
< FFN to evaluate whether it is an n-back match.  The model responds “match” if the FFN detects a match; otherwise, it
< either responds “non-match” or, with a fixed probability (hazard rate), it uses the current stimulus and temporal
< context to retrieve another sample from EM and repeat the evaluation.</p>
< <p>This model is an example of proposed interactions between working memory (e.g., in neocortex) and episodic memory
< e.g., in hippocampus and/or cerebellum) in the performance of tasks demanding of sequential processing and control,
< and along the lines of models emerging machine learning that augment the use of recurrent neural networks (e.g., long
< short-term memory mechanisms; LSTMs) for active memory and control with an external memory capable of rapid storage
< and content-based retrieval, such as the Neural Turing Machine (NTN;
< <a class="reference external" href="https://arxiv.org/abs/1410.5401">Graves et al., 2016</a>), Episodic Planning Networks (EPN;
< <a class="reference external" href="https://arxiv.org/abs/2006.03662">Ritter et al., 2020</a>), and Emergent Symbols through Binding Networks (ESBN;
< <a class="reference external" href="https://arxiv.org/abs/2012.14601">Webb et al., 2021</a>).</p>
< <p>The script respectively, to construct, train and run the model:</p>
< <ul class="simple">
< <li><p>construct_model(args):
< takes as arguments parameters used to construct the model; for convenience, defaults are defined toward the top
< of the script (see “Construction parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>train_network(args)
< takes as arguments the feedforward neural network Composition (FFN_COMPOSITION) and number of epochs to train.
< Note: learning_rate is set at construction (which can be specified using LEARNING_RATE under “Training parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>run_model()
< takes as arguments the drift rate in the temporal context vector to be applied on each trial,
< and the number of trials to execute, as well as reporting and animation specifications
< (see “Execution parameters”).</p></li>
< </ul>
< <p>The default parameters are ones that have been fit to empirical data concerning human performance
< (taken from <a class="reference external" href="https://psycnet.apa.org/record/2007-06096-010?doi=1">Kane et al., 2007</a>).</p>
< </section>
< <section id="the-model">
< <h2>The Model<a class="headerlink" href="#the-model" title="Permalink to this headline">¶</a></h2>
< <p>The models is composed of two <a class="reference internal" href="Composition.html"><span class="doc">Compositions</span></a>: an outer one that contains the full model (nback_model),
< and an <a class="reference internal" href="AutodiffComposition.html"><span class="doc">AutodiffComposition</span></a> (ffn), nested within nback_model (see red box in Figure), that implements the
< feedforward neural network (ffn).</p>
< <section id="nback-model">
< <h3>nback_model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h3>
< <p>This contains three input Mechanisms (</p>
< <p>Both of these are constructed in the construct_model function.
< The ffn Composition is trained use</p>
< <figure class="align-left" id="nback-fig">
< <img alt="N-Back Model Animation" src="_images/N-Back_Model_movie.gif" />
< </figure>
< </section>
< </section>
< <section id="training">
< <h2>Training<a class="headerlink" href="#training" title="Permalink to this headline">¶</a></h2>
< </section>
< <section id="execution">
< <h2>Execution<a class="headerlink" href="#execution" title="Permalink to this headline">¶</a></h2>
< <p>Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code>
< .. Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code></p>
< </section>
---
>   <section id="nback-model">
> <h1>Nback Model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h1>
> <p><strong>“When Working Memory is Just Working, Not Memory”</strong> (<a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al., 2022</a>)</p>
> <p>Script: <a class="reference download internal" download="" href="_downloads/3ca6d8ab14ce9c785a6b5c1b1d60a955/nback.py"><code class="xref download docutils literal notranslate"><span class="pre">nback.py</span></code></a></p>
> <p>Example of Jupyter Notebook: <a class="reference internal" href="BeukersNBackModel_NB.html"><span class="doc">Nback Model Notebook</span></a></p>
> <p>(Instructions for running the actual notebook can be found <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">here</span></a>, replacing <code class="docutils literal notranslate"><span class="pre">nback_nb</span></code> for <code class="docutils literal notranslate"><span class="pre">tutorial</span></code>)</p>
> <div class="toctree-wrapper compound">
> </div>
276a222,230
>     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
>       
>         <a href="ContributorsGuide.html" class="btn btn-neutral float-right" title="Contributors Guide" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
>       
>       
>         <a href="BustamanteStroopXORLVOCModel.html" class="btn btn-neutral" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
>       
>     </div>
>   
305,314c259
< <li><a class="reference internal" href="#">N-Back Model
...

See CI logs for the full diff.

[github-actions]

This PR causes the following changes to the html docs (ubuntu-latest-3.7-x64):

diff -r docs-base/AutodiffComposition.html docs-head/AutodiffComposition.html
293,294c293,294
< This <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a> provides a summary and comparison of these different modes of execution,
< that are described in greater detail below.</p>
---
> These are each described in greater detail below, and summarized in this <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a>
> which provides a comparison of the different modes of execution for an AutodiffComposition and standard <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a>.</p>
325c325
< executing using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
---
> executed using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
327a328,341
> <div class="admonition warning">
> <p class="admonition-title">Warning</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.LLVM" title="psyneulink.core.llvm.__init__.ExecutionMode.LLVM"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.LLVM</span></code></a> or <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> in the learn() method of a standard
> <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a> causes an error.</p></li>
> </ul>
> </div>
> <div class="admonition note">
> <p class="admonition-title">Note</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.Python" title="psyneulink.core.llvm.__init__.ExecutionMode.Python"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.Python</span></code></a> in the learn() method of a <a class="reference internal" href="#"><span class="doc">AutodiffComposition</span></a> is treated as a
> synonym of <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> (see table).</p></li>
> </ul>
> </div>
415a430,452
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
> <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
> <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
> arguments from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch optimizer function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
> <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
> <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch loss function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
418c455
< <dd><p>tracks the average for each weight update (i.e. each minibatch)</p>
---
> <dd><p>tracks the average loss after each weight update (i.e. each minibatch) during learning.</p>
427,430c464,466
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
< <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
< <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
< arguments from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights">
> <span class="sig-name descname"><span class="pre">last_saved_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file to which weights were last saved.</p>
433c469
< <dd class="field-odd"><p>PyTorch optimizer function</p>
---
> <dd class="field-odd"><p>path</p>
439,441c475,477
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
< <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
< <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights">
> <span class="sig-name descname"><span class="pre">last_loaded_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file from which weights were last loaded.</p>
444c480
< <dd class="field-odd"><p>PyTorch loss function</p>
---
> <dd class="field-odd"><p>path</p>
457c493
< <dd><p>Updates parameters based on trials run since last update.</p>
---
> <dd><p>Updates parameters (weights) based on trials run since last update.</p>
636c672
< <dd><p>Loads all weights matrices for all MappingProjections in the AutodiffComposition from file
---
> <dd><p>Loads all weight matrices for all MappingProjections in the AutodiffComposition from file
diff -r docs-base/BasicsAndPrimer.html docs-head/BasicsAndPrimer.html
259c259
< <span id="basicsandprimer-grandview-figure"></span><img alt="_static/BasicsAndPrimer_GrandView_fig.svg" src="_static/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
---
> <span id="basicsandprimer-grandview-figure"></span><img alt="_images/BasicsAndPrimer_GrandView_fig.svg" src="_images/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
269,270c269,270
< <code class="xref any docutils literal notranslate"><span class="pre">tutorial</span></code> provides additional introductory material for those who are newer to computational modeling, as well as a
< more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
---
> <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">tutorial</span></a> provides additional introductory material for those who are newer to computational modeling,
> as well as a more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
diff -r docs-base/BeukersNBackModel.html docs-head/BeukersNBackModel.html
13c13
<   <title>N-Back Model (Beukers et al., 2022) &mdash; PsyNeuLink 0.0.0.0 documentation</title>
---
>   <title>Nback Model &mdash; PsyNeuLink 0.0.0.0 documentation</title>
35c35,37
<     <link rel="search" title="Search" href="search.html" /> 
---
>     <link rel="search" title="Search" href="search.html" />
>     <link rel="next" title="Contributors Guide" href="ContributorsGuide.html" />
>     <link rel="prev" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" href="BustamanteStroopXORLVOCModel.html" /> 
129c131
<               <ul>
---
>               <ul class="current">
134c136
< <li class="toctree-l1"><a class="reference internal" href="Library.html">Library</a></li>
---
> <li class="toctree-l1 current"><a class="reference internal" href="Library.html">Library</a></li>
177c179,183
<       <li>N-Back Model (Beukers et al., 2022)</li>
---
>           <li><a href="Library.html">Library</a> &gt;</li>
>         
>           <li><a href="Models.html">Models</a> &gt;</li>
>         
>       <li>Nback Model</li>
200,268c206,213
<   <section id="n-back-model-beukers-et-al-2022">
< <h1>N-Back Model (Beukers et al., 2022)<a class="headerlink" href="#n-back-model-beukers-et-al-2022" title="Permalink to this headline">¶</a></h1>
< <p><a class="reference external" href="https://psyarxiv.com/jtw5p">“When Working Memory is Just Working, Not Memory”</a></p>
< <section id="overview">
< <h2>Overview<a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h2>
< <p>This implements a model of the <a class="reference external" href="https://en.wikipedia.org/wiki/N-back#Neurobiology_of_n-back_task">N-back task</a>
< described in <a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al. (2022)</a>.  The model uses a simple implementation of episodic
< memory (EM, as a form of content-retrieval memory) to store previous stimuli along with the temporal context in which
< they occured, and a feedforward neural network (FFN)to evaluate whether the current stimulus is a match to the n’th
< preceding stimulus (nback-level)retrieved from episodic memory.  The temporal context is provided by a randomly
< drifting high dimensional vector that maintains a constant norm (i.e., drifts on a sphere).  The FFN is
< trained, given an n-back level of <em>n</em>, to identify when the current stimulus matches one stored in EM
< with a temporal context vector that differs by an amount corresponding to <em>n</em> time steps of drift.  During n-back
< performance, the model encodes the current stimulus and temporal context, retrieves an item from EM that matches the
< current stimulus, weighted by the similarity of its temporal context vector (i.e., most recent), and then uses the
< FFN to evaluate whether it is an n-back match.  The model responds “match” if the FFN detects a match; otherwise, it
< either responds “non-match” or, with a fixed probability (hazard rate), it uses the current stimulus and temporal
< context to retrieve another sample from EM and repeat the evaluation.</p>
< <p>This model is an example of proposed interactions between working memory (e.g., in neocortex) and episodic memory
< e.g., in hippocampus and/or cerebellum) in the performance of tasks demanding of sequential processing and control,
< and along the lines of models emerging machine learning that augment the use of recurrent neural networks (e.g., long
< short-term memory mechanisms; LSTMs) for active memory and control with an external memory capable of rapid storage
< and content-based retrieval, such as the Neural Turing Machine (NTN;
< <a class="reference external" href="https://arxiv.org/abs/1410.5401">Graves et al., 2016</a>), Episodic Planning Networks (EPN;
< <a class="reference external" href="https://arxiv.org/abs/2006.03662">Ritter et al., 2020</a>), and Emergent Symbols through Binding Networks (ESBN;
< <a class="reference external" href="https://arxiv.org/abs/2012.14601">Webb et al., 2021</a>).</p>
< <p>The script respectively, to construct, train and run the model:</p>
< <ul class="simple">
< <li><p>construct_model(args):
< takes as arguments parameters used to construct the model; for convenience, defaults are defined toward the top
< of the script (see “Construction parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>train_network(args)
< takes as arguments the feedforward neural network Composition (FFN_COMPOSITION) and number of epochs to train.
< Note: learning_rate is set at construction (which can be specified using LEARNING_RATE under “Training parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>run_model()
< takes as arguments the drift rate in the temporal context vector to be applied on each trial,
< and the number of trials to execute, as well as reporting and animation specifications
< (see “Execution parameters”).</p></li>
< </ul>
< <p>The default parameters are ones that have been fit to empirical data concerning human performance
< (taken from <a class="reference external" href="https://psycnet.apa.org/record/2007-06096-010?doi=1">Kane et al., 2007</a>).</p>
< </section>
< <section id="the-model">
< <h2>The Model<a class="headerlink" href="#the-model" title="Permalink to this headline">¶</a></h2>
< <p>The models is composed of two <a class="reference internal" href="Composition.html"><span class="doc">Compositions</span></a>: an outer one that contains the full model (nback_model),
< and an <a class="reference internal" href="AutodiffComposition.html"><span class="doc">AutodiffComposition</span></a> (ffn), nested within nback_model (see red box in Figure), that implements the
< feedforward neural network (ffn).</p>
< <section id="nback-model">
< <h3>nback_model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h3>
< <p>This contains three input Mechanisms (</p>
< <p>Both of these are constructed in the construct_model function.
< The ffn Composition is trained use</p>
< <figure class="align-left" id="nback-fig">
< <img alt="N-Back Model Animation" src="_images/N-Back_Model_movie.gif" />
< </figure>
< </section>
< </section>
< <section id="training">
< <h2>Training<a class="headerlink" href="#training" title="Permalink to this headline">¶</a></h2>
< </section>
< <section id="execution">
< <h2>Execution<a class="headerlink" href="#execution" title="Permalink to this headline">¶</a></h2>
< <p>Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code>
< .. Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code></p>
< </section>
---
>   <section id="nback-model">
> <h1>Nback Model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h1>
> <p><strong>“When Working Memory is Just Working, Not Memory”</strong> (<a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al., 2022</a>)</p>
> <p>Script: <a class="reference download internal" download="" href="_downloads/3ca6d8ab14ce9c785a6b5c1b1d60a955/nback.py"><code class="xref download docutils literal notranslate"><span class="pre">nback.py</span></code></a></p>
> <p>Example of Jupyter Notebook: <a class="reference internal" href="BeukersNBackModel_NB.html"><span class="doc">Nback Model Notebook</span></a></p>
> <p>(Instructions for running the actual notebook can be found <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">here</span></a>, replacing <code class="docutils literal notranslate"><span class="pre">nback_nb</span></code> for <code class="docutils literal notranslate"><span class="pre">tutorial</span></code>)</p>
> <div class="toctree-wrapper compound">
> </div>
276a222,230
>     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
>       
>         <a href="ContributorsGuide.html" class="btn btn-neutral float-right" title="Contributors Guide" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
>       
>       
>         <a href="BustamanteStroopXORLVOCModel.html" class="btn btn-neutral" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
>       
>     </div>
>   
305,314c259
< <li><a class="reference internal" href="#">N-Back Model
...

See CI logs for the full diff.

[github-actions]

This PR causes the following changes to the html docs (ubuntu-latest-3.7-x64):

diff -r docs-base/AutodiffComposition.html docs-head/AutodiffComposition.html
293,294c293,294
< This <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a> provides a summary and comparison of these different modes of execution,
< that are described in greater detail below.</p>
---
> These are each described in greater detail below, and summarized in this <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a>
> which provides a comparison of the different modes of execution for an AutodiffComposition and standard <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a>.</p>
325c325
< executing using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
---
> executed using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
327a328,341
> <div class="admonition warning">
> <p class="admonition-title">Warning</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.LLVM" title="psyneulink.core.llvm.__init__.ExecutionMode.LLVM"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.LLVM</span></code></a> or <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> in the learn() method of a standard
> <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a> causes an error.</p></li>
> </ul>
> </div>
> <div class="admonition note">
> <p class="admonition-title">Note</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.Python" title="psyneulink.core.llvm.__init__.ExecutionMode.Python"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.Python</span></code></a> in the learn() method of a <a class="reference internal" href="#"><span class="doc">AutodiffComposition</span></a> is treated as a
> synonym of <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> (see table).</p></li>
> </ul>
> </div>
415a430,452
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
> <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
> <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
> arguments from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch optimizer function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
> <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
> <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch loss function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
418c455
< <dd><p>tracks the average for each weight update (i.e. each minibatch)</p>
---
> <dd><p>tracks the average loss after each weight update (i.e. each minibatch) during learning.</p>
427,430c464,466
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
< <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
< <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
< arguments from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights">
> <span class="sig-name descname"><span class="pre">last_saved_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file to which weights were last saved.</p>
433c469
< <dd class="field-odd"><p>PyTorch optimizer function</p>
---
> <dd class="field-odd"><p>path</p>
439,441c475,477
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
< <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
< <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights">
> <span class="sig-name descname"><span class="pre">last_loaded_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file from which weights were last loaded.</p>
444c480
< <dd class="field-odd"><p>PyTorch loss function</p>
---
> <dd class="field-odd"><p>path</p>
457c493
< <dd><p>Updates parameters based on trials run since last update.</p>
---
> <dd><p>Updates parameters (weights) based on trials run since last update.</p>
636c672
< <dd><p>Loads all weights matrices for all MappingProjections in the AutodiffComposition from file
---
> <dd><p>Loads all weight matrices for all MappingProjections in the AutodiffComposition from file
diff -r docs-base/BasicsAndPrimer.html docs-head/BasicsAndPrimer.html
259c259
< <span id="basicsandprimer-grandview-figure"></span><img alt="_static/BasicsAndPrimer_GrandView_fig.svg" src="_static/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
---
> <span id="basicsandprimer-grandview-figure"></span><img alt="_images/BasicsAndPrimer_GrandView_fig.svg" src="_images/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
269,270c269,270
< <code class="xref any docutils literal notranslate"><span class="pre">tutorial</span></code> provides additional introductory material for those who are newer to computational modeling, as well as a
< more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
---
> <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">tutorial</span></a> provides additional introductory material for those who are newer to computational modeling,
> as well as a more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
diff -r docs-base/BeukersNBackModel.html docs-head/BeukersNBackModel.html
13c13
<   <title>N-Back Model (Beukers et al., 2022) &mdash; PsyNeuLink 0.0.0.0 documentation</title>
---
>   <title>Nback Model &mdash; PsyNeuLink 0.0.0.0 documentation</title>
35c35,37
<     <link rel="search" title="Search" href="search.html" /> 
---
>     <link rel="search" title="Search" href="search.html" />
>     <link rel="next" title="Contributors Guide" href="ContributorsGuide.html" />
>     <link rel="prev" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" href="BustamanteStroopXORLVOCModel.html" /> 
129c131
<               <ul>
---
>               <ul class="current">
134c136
< <li class="toctree-l1"><a class="reference internal" href="Library.html">Library</a></li>
---
> <li class="toctree-l1 current"><a class="reference internal" href="Library.html">Library</a></li>
177c179,183
<       <li>N-Back Model (Beukers et al., 2022)</li>
---
>           <li><a href="Library.html">Library</a> &gt;</li>
>         
>           <li><a href="Models.html">Models</a> &gt;</li>
>         
>       <li>Nback Model</li>
200,268c206,213
<   <section id="n-back-model-beukers-et-al-2022">
< <h1>N-Back Model (Beukers et al., 2022)<a class="headerlink" href="#n-back-model-beukers-et-al-2022" title="Permalink to this headline">¶</a></h1>
< <p><a class="reference external" href="https://psyarxiv.com/jtw5p">“When Working Memory is Just Working, Not Memory”</a></p>
< <section id="overview">
< <h2>Overview<a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h2>
< <p>This implements a model of the <a class="reference external" href="https://en.wikipedia.org/wiki/N-back#Neurobiology_of_n-back_task">N-back task</a>
< described in <a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al. (2022)</a>.  The model uses a simple implementation of episodic
< memory (EM, as a form of content-retrieval memory) to store previous stimuli along with the temporal context in which
< they occured, and a feedforward neural network (FFN)to evaluate whether the current stimulus is a match to the n’th
< preceding stimulus (nback-level)retrieved from episodic memory.  The temporal context is provided by a randomly
< drifting high dimensional vector that maintains a constant norm (i.e., drifts on a sphere).  The FFN is
< trained, given an n-back level of <em>n</em>, to identify when the current stimulus matches one stored in EM
< with a temporal context vector that differs by an amount corresponding to <em>n</em> time steps of drift.  During n-back
< performance, the model encodes the current stimulus and temporal context, retrieves an item from EM that matches the
< current stimulus, weighted by the similarity of its temporal context vector (i.e., most recent), and then uses the
< FFN to evaluate whether it is an n-back match.  The model responds “match” if the FFN detects a match; otherwise, it
< either responds “non-match” or, with a fixed probability (hazard rate), it uses the current stimulus and temporal
< context to retrieve another sample from EM and repeat the evaluation.</p>
< <p>This model is an example of proposed interactions between working memory (e.g., in neocortex) and episodic memory
< e.g., in hippocampus and/or cerebellum) in the performance of tasks demanding of sequential processing and control,
< and along the lines of models emerging machine learning that augment the use of recurrent neural networks (e.g., long
< short-term memory mechanisms; LSTMs) for active memory and control with an external memory capable of rapid storage
< and content-based retrieval, such as the Neural Turing Machine (NTN;
< <a class="reference external" href="https://arxiv.org/abs/1410.5401">Graves et al., 2016</a>), Episodic Planning Networks (EPN;
< <a class="reference external" href="https://arxiv.org/abs/2006.03662">Ritter et al., 2020</a>), and Emergent Symbols through Binding Networks (ESBN;
< <a class="reference external" href="https://arxiv.org/abs/2012.14601">Webb et al., 2021</a>).</p>
< <p>The script respectively, to construct, train and run the model:</p>
< <ul class="simple">
< <li><p>construct_model(args):
< takes as arguments parameters used to construct the model; for convenience, defaults are defined toward the top
< of the script (see “Construction parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>train_network(args)
< takes as arguments the feedforward neural network Composition (FFN_COMPOSITION) and number of epochs to train.
< Note: learning_rate is set at construction (which can be specified using LEARNING_RATE under “Training parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>run_model()
< takes as arguments the drift rate in the temporal context vector to be applied on each trial,
< and the number of trials to execute, as well as reporting and animation specifications
< (see “Execution parameters”).</p></li>
< </ul>
< <p>The default parameters are ones that have been fit to empirical data concerning human performance
< (taken from <a class="reference external" href="https://psycnet.apa.org/record/2007-06096-010?doi=1">Kane et al., 2007</a>).</p>
< </section>
< <section id="the-model">
< <h2>The Model<a class="headerlink" href="#the-model" title="Permalink to this headline">¶</a></h2>
< <p>The models is composed of two <a class="reference internal" href="Composition.html"><span class="doc">Compositions</span></a>: an outer one that contains the full model (nback_model),
< and an <a class="reference internal" href="AutodiffComposition.html"><span class="doc">AutodiffComposition</span></a> (ffn), nested within nback_model (see red box in Figure), that implements the
< feedforward neural network (ffn).</p>
< <section id="nback-model">
< <h3>nback_model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h3>
< <p>This contains three input Mechanisms (</p>
< <p>Both of these are constructed in the construct_model function.
< The ffn Composition is trained use</p>
< <figure class="align-left" id="nback-fig">
< <img alt="N-Back Model Animation" src="_images/N-Back_Model_movie.gif" />
< </figure>
< </section>
< </section>
< <section id="training">
< <h2>Training<a class="headerlink" href="#training" title="Permalink to this headline">¶</a></h2>
< </section>
< <section id="execution">
< <h2>Execution<a class="headerlink" href="#execution" title="Permalink to this headline">¶</a></h2>
< <p>Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code>
< .. Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code></p>
< </section>
---
>   <section id="nback-model">
> <h1>Nback Model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h1>
> <p><strong>“When Working Memory is Just Working, Not Memory”</strong> (<a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al., 2022</a>)</p>
> <p>Script: <a class="reference download internal" download="" href="_downloads/3ca6d8ab14ce9c785a6b5c1b1d60a955/nback.py"><code class="xref download docutils literal notranslate"><span class="pre">nback.py</span></code></a></p>
> <p>Example of Jupyter Notebook: <a class="reference internal" href="BeukersNBackModel_NB.html"><span class="doc">Nback Model Notebook</span></a></p>
> <p>(Instructions for running the actual notebook can be found <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">here</span></a>, replacing <code class="docutils literal notranslate"><span class="pre">nback_nb</span></code> for <code class="docutils literal notranslate"><span class="pre">tutorial</span></code>)</p>
> <div class="toctree-wrapper compound">
> </div>
276a222,230
>     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
>       
>         <a href="ContributorsGuide.html" class="btn btn-neutral float-right" title="Contributors Guide" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
>       
>       
>         <a href="BustamanteStroopXORLVOCModel.html" class="btn btn-neutral" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
>       
>     </div>
>   
305,314c259
< <li><a class="reference internal" href="#">N-Back Model
...

See CI logs for the full diff.

[github-actions]

This PR causes the following changes to the html docs (ubuntu-latest-3.7-x64):

diff -r docs-base/AutodiffComposition.html docs-head/AutodiffComposition.html
293,294c293,294
< This <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a> provides a summary and comparison of these different modes of execution,
< that are described in greater detail below.</p>
---
> These are each described in greater detail below, and summarized in this <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a>
> which provides a comparison of the different modes of execution for an AutodiffComposition and standard <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a>.</p>
325c325
< executing using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
---
> executed using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
327a328,341
> <div class="admonition warning">
> <p class="admonition-title">Warning</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.LLVM" title="psyneulink.core.llvm.__init__.ExecutionMode.LLVM"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.LLVM</span></code></a> or <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> in the learn() method of a standard
> <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a> causes an error.</p></li>
> </ul>
> </div>
> <div class="admonition note">
> <p class="admonition-title">Note</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.Python" title="psyneulink.core.llvm.__init__.ExecutionMode.Python"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.Python</span></code></a> in the learn() method of a <a class="reference internal" href="#"><span class="doc">AutodiffComposition</span></a> is treated as a
> synonym of <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> (see table).</p></li>
> </ul>
> </div>
415a430,452
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
> <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
> <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
> arguments from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch optimizer function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
> <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
> <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch loss function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
418c455
< <dd><p>tracks the average for each weight update (i.e. each minibatch)</p>
---
> <dd><p>tracks the average loss after each weight update (i.e. each minibatch) during learning.</p>
427,430c464,466
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
< <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
< <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
< arguments from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights">
> <span class="sig-name descname"><span class="pre">last_saved_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file to which weights were last saved.</p>
433c469
< <dd class="field-odd"><p>PyTorch optimizer function</p>
---
> <dd class="field-odd"><p>path</p>
439,441c475,477
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
< <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
< <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights">
> <span class="sig-name descname"><span class="pre">last_loaded_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file from which weights were last loaded.</p>
444c480
< <dd class="field-odd"><p>PyTorch loss function</p>
---
> <dd class="field-odd"><p>path</p>
457c493
< <dd><p>Updates parameters based on trials run since last update.</p>
---
> <dd><p>Updates parameters (weights) based on trials run since last update.</p>
636c672
< <dd><p>Loads all weights matrices for all MappingProjections in the AutodiffComposition from file
---
> <dd><p>Loads all weight matrices for all MappingProjections in the AutodiffComposition from file
diff -r docs-base/BasicsAndPrimer.html docs-head/BasicsAndPrimer.html
259c259
< <span id="basicsandprimer-grandview-figure"></span><img alt="_static/BasicsAndPrimer_GrandView_fig.svg" src="_static/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
---
> <span id="basicsandprimer-grandview-figure"></span><img alt="_images/BasicsAndPrimer_GrandView_fig.svg" src="_images/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
269,270c269,270
< <code class="xref any docutils literal notranslate"><span class="pre">tutorial</span></code> provides additional introductory material for those who are newer to computational modeling, as well as a
< more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
---
> <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">tutorial</span></a> provides additional introductory material for those who are newer to computational modeling,
> as well as a more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
diff -r docs-base/BeukersNBackModel.html docs-head/BeukersNBackModel.html
13c13
<   <title>N-Back Model (Beukers et al., 2022) &mdash; PsyNeuLink 0.0.0.0 documentation</title>
---
>   <title>Nback Model &mdash; PsyNeuLink 0.0.0.0 documentation</title>
35c35,37
<     <link rel="search" title="Search" href="search.html" /> 
---
>     <link rel="search" title="Search" href="search.html" />
>     <link rel="next" title="Contributors Guide" href="ContributorsGuide.html" />
>     <link rel="prev" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" href="BustamanteStroopXORLVOCModel.html" /> 
129c131
<               <ul>
---
>               <ul class="current">
134c136
< <li class="toctree-l1"><a class="reference internal" href="Library.html">Library</a></li>
---
> <li class="toctree-l1 current"><a class="reference internal" href="Library.html">Library</a></li>
177c179,183
<       <li>N-Back Model (Beukers et al., 2022)</li>
---
>           <li><a href="Library.html">Library</a> &gt;</li>
>         
>           <li><a href="Models.html">Models</a> &gt;</li>
>         
>       <li>Nback Model</li>
200,268c206,213
<   <section id="n-back-model-beukers-et-al-2022">
< <h1>N-Back Model (Beukers et al., 2022)<a class="headerlink" href="#n-back-model-beukers-et-al-2022" title="Permalink to this headline">¶</a></h1>
< <p><a class="reference external" href="https://psyarxiv.com/jtw5p">“When Working Memory is Just Working, Not Memory”</a></p>
< <section id="overview">
< <h2>Overview<a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h2>
< <p>This implements a model of the <a class="reference external" href="https://en.wikipedia.org/wiki/N-back#Neurobiology_of_n-back_task">N-back task</a>
< described in <a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al. (2022)</a>.  The model uses a simple implementation of episodic
< memory (EM, as a form of content-retrieval memory) to store previous stimuli along with the temporal context in which
< they occured, and a feedforward neural network (FFN)to evaluate whether the current stimulus is a match to the n’th
< preceding stimulus (nback-level)retrieved from episodic memory.  The temporal context is provided by a randomly
< drifting high dimensional vector that maintains a constant norm (i.e., drifts on a sphere).  The FFN is
< trained, given an n-back level of <em>n</em>, to identify when the current stimulus matches one stored in EM
< with a temporal context vector that differs by an amount corresponding to <em>n</em> time steps of drift.  During n-back
< performance, the model encodes the current stimulus and temporal context, retrieves an item from EM that matches the
< current stimulus, weighted by the similarity of its temporal context vector (i.e., most recent), and then uses the
< FFN to evaluate whether it is an n-back match.  The model responds “match” if the FFN detects a match; otherwise, it
< either responds “non-match” or, with a fixed probability (hazard rate), it uses the current stimulus and temporal
< context to retrieve another sample from EM and repeat the evaluation.</p>
< <p>This model is an example of proposed interactions between working memory (e.g., in neocortex) and episodic memory
< e.g., in hippocampus and/or cerebellum) in the performance of tasks demanding of sequential processing and control,
< and along the lines of models emerging machine learning that augment the use of recurrent neural networks (e.g., long
< short-term memory mechanisms; LSTMs) for active memory and control with an external memory capable of rapid storage
< and content-based retrieval, such as the Neural Turing Machine (NTN;
< <a class="reference external" href="https://arxiv.org/abs/1410.5401">Graves et al., 2016</a>), Episodic Planning Networks (EPN;
< <a class="reference external" href="https://arxiv.org/abs/2006.03662">Ritter et al., 2020</a>), and Emergent Symbols through Binding Networks (ESBN;
< <a class="reference external" href="https://arxiv.org/abs/2012.14601">Webb et al., 2021</a>).</p>
< <p>The script respectively, to construct, train and run the model:</p>
< <ul class="simple">
< <li><p>construct_model(args):
< takes as arguments parameters used to construct the model; for convenience, defaults are defined toward the top
< of the script (see “Construction parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>train_network(args)
< takes as arguments the feedforward neural network Composition (FFN_COMPOSITION) and number of epochs to train.
< Note: learning_rate is set at construction (which can be specified using LEARNING_RATE under “Training parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>run_model()
< takes as arguments the drift rate in the temporal context vector to be applied on each trial,
< and the number of trials to execute, as well as reporting and animation specifications
< (see “Execution parameters”).</p></li>
< </ul>
< <p>The default parameters are ones that have been fit to empirical data concerning human performance
< (taken from <a class="reference external" href="https://psycnet.apa.org/record/2007-06096-010?doi=1">Kane et al., 2007</a>).</p>
< </section>
< <section id="the-model">
< <h2>The Model<a class="headerlink" href="#the-model" title="Permalink to this headline">¶</a></h2>
< <p>The models is composed of two <a class="reference internal" href="Composition.html"><span class="doc">Compositions</span></a>: an outer one that contains the full model (nback_model),
< and an <a class="reference internal" href="AutodiffComposition.html"><span class="doc">AutodiffComposition</span></a> (ffn), nested within nback_model (see red box in Figure), that implements the
< feedforward neural network (ffn).</p>
< <section id="nback-model">
< <h3>nback_model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h3>
< <p>This contains three input Mechanisms (</p>
< <p>Both of these are constructed in the construct_model function.
< The ffn Composition is trained use</p>
< <figure class="align-left" id="nback-fig">
< <img alt="N-Back Model Animation" src="_images/N-Back_Model_movie.gif" />
< </figure>
< </section>
< </section>
< <section id="training">
< <h2>Training<a class="headerlink" href="#training" title="Permalink to this headline">¶</a></h2>
< </section>
< <section id="execution">
< <h2>Execution<a class="headerlink" href="#execution" title="Permalink to this headline">¶</a></h2>
< <p>Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code>
< .. Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code></p>
< </section>
---
>   <section id="nback-model">
> <h1>Nback Model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h1>
> <p><strong>“When Working Memory is Just Working, Not Memory”</strong> (<a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al., 2022</a>)</p>
> <p>Script: <a class="reference download internal" download="" href="_downloads/3ca6d8ab14ce9c785a6b5c1b1d60a955/nback.py"><code class="xref download docutils literal notranslate"><span class="pre">nback.py</span></code></a></p>
> <p>Example of Jupyter Notebook: <a class="reference internal" href="BeukersNBackModel_NB.html"><span class="doc">Nback Model Notebook</span></a></p>
> <p>(Instructions for running the actual notebook can be found <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">here</span></a>, replacing <code class="docutils literal notranslate"><span class="pre">nback_nb</span></code> for <code class="docutils literal notranslate"><span class="pre">tutorial</span></code>)</p>
> <div class="toctree-wrapper compound">
> </div>
276a222,230
>     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
>       
>         <a href="ContributorsGuide.html" class="btn btn-neutral float-right" title="Contributors Guide" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
>       
>       
>         <a href="BustamanteStroopXORLVOCModel.html" class="btn btn-neutral" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
>       
>     </div>
>   
305,314c259
< <li><a class="reference internal" href="#">N-Back Model
...

See CI logs for the full diff.

[github-actions]

This PR causes the following changes to the html docs (ubuntu-latest-3.7-x64):

diff -r docs-base/AutodiffComposition.html docs-head/AutodiffComposition.html
293,294c293,294
< This <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a> provides a summary and comparison of these different modes of execution,
< that are described in greater detail below.</p>
---
> These are each described in greater detail below, and summarized in this <a class="reference internal" href="Composition.html#composition-compilation-table"><span class="std std-ref">table</span></a>
> which provides a comparison of the different modes of execution for an AutodiffComposition and standard <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a>.</p>
325c325
< executing using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
---
> executed using the <em>Python</em> interpreter (and not PyTorch);  this is so that any modulation can take effect
327a328,341
> <div class="admonition warning">
> <p class="admonition-title">Warning</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.LLVM" title="psyneulink.core.llvm.__init__.ExecutionMode.LLVM"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.LLVM</span></code></a> or <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> in the learn() method of a standard
> <a class="reference internal" href="Composition.html"><span class="doc">Composition</span></a> causes an error.</p></li>
> </ul>
> </div>
> <div class="admonition note">
> <p class="admonition-title">Note</p>
> <ul class="simple">
> <li><p>Specifying <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.Python" title="psyneulink.core.llvm.__init__.ExecutionMode.Python"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.Python</span></code></a> in the learn() method of a <a class="reference internal" href="#"><span class="doc">AutodiffComposition</span></a> is treated as a
> synonym of <a class="reference internal" href="LLVM.html#psyneulink.core.llvm.__init__.ExecutionMode.PyTorch" title="psyneulink.core.llvm.__init__.ExecutionMode.PyTorch"><code class="xref any py py-attr docutils literal notranslate"><span class="pre">ExecutionMode.PyTorch</span></code></a> (see table).</p></li>
> </ul>
> </div>
415a430,452
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
> <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
> <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
> arguments from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch optimizer function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
> <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
> <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
> <dl class="field-list simple">
> <dt class="field-odd">Type</dt>
> <dd class="field-odd"><p>PyTorch loss function</p>
> </dd>
> </dl>
> </dd></dl>
> 
> <dl class="py attribute">
418c455
< <dd><p>tracks the average for each weight update (i.e. each minibatch)</p>
---
> <dd><p>tracks the average loss after each weight update (i.e. each minibatch) during learning.</p>
427,430c464,466
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer">
< <span class="sig-name descname"><span class="pre">optimizer</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.optimizer" title="Permalink to this definition">¶</a></dt>
< <dd><p>the optimizer used for training. Depends on the <strong>optimizer_type</strong>, <strong>learning_rate</strong>, and <strong>weight_decay</strong>
< arguments from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights">
> <span class="sig-name descname"><span class="pre">last_saved_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_saved_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file to which weights were last saved.</p>
433c469
< <dd class="field-odd"><p>PyTorch optimizer function</p>
---
> <dd class="field-odd"><p>path</p>
439,441c475,477
< <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss">
< <span class="sig-name descname"><span class="pre">loss</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.loss" title="Permalink to this definition">¶</a></dt>
< <dd><p>the loss function used for training. Depends on the <strong>loss_spec</strong> argument from initialization.</p>
---
> <dt class="sig sig-object py" id="psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights">
> <span class="sig-name descname"><span class="pre">last_loaded_weights</span></span><a class="headerlink" href="#psyneulink.library.compositions.autodiffcomposition.AutodiffComposition.last_loaded_weights" title="Permalink to this definition">¶</a></dt>
> <dd><p>path for file from which weights were last loaded.</p>
444c480
< <dd class="field-odd"><p>PyTorch loss function</p>
---
> <dd class="field-odd"><p>path</p>
457c493
< <dd><p>Updates parameters based on trials run since last update.</p>
---
> <dd><p>Updates parameters (weights) based on trials run since last update.</p>
636c672
< <dd><p>Loads all weights matrices for all MappingProjections in the AutodiffComposition from file
---
> <dd><p>Loads all weight matrices for all MappingProjections in the AutodiffComposition from file
diff -r docs-base/BasicsAndPrimer.html docs-head/BasicsAndPrimer.html
259c259
< <span id="basicsandprimer-grandview-figure"></span><img alt="_static/BasicsAndPrimer_GrandView_fig.svg" src="_static/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
---
> <span id="basicsandprimer-grandview-figure"></span><img alt="_images/BasicsAndPrimer_GrandView_fig.svg" src="_images/BasicsAndPrimer_GrandView_fig.svg" /><figcaption>
269,270c269,270
< <code class="xref any docutils literal notranslate"><span class="pre">tutorial</span></code> provides additional introductory material for those who are newer to computational modeling, as well as a
< more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
---
> <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">tutorial</span></a> provides additional introductory material for those who are newer to computational modeling,
> as well as a more detailed and comprehensive introduction to the use of PsyNeuLink.</p>
diff -r docs-base/BeukersNBackModel.html docs-head/BeukersNBackModel.html
13c13
<   <title>N-Back Model (Beukers et al., 2022) &mdash; PsyNeuLink 0.0.0.0 documentation</title>
---
>   <title>Nback Model &mdash; PsyNeuLink 0.0.0.0 documentation</title>
35c35,37
<     <link rel="search" title="Search" href="search.html" /> 
---
>     <link rel="search" title="Search" href="search.html" />
>     <link rel="next" title="Contributors Guide" href="ContributorsGuide.html" />
>     <link rel="prev" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" href="BustamanteStroopXORLVOCModel.html" /> 
129c131
<               <ul>
---
>               <ul class="current">
134c136
< <li class="toctree-l1"><a class="reference internal" href="Library.html">Library</a></li>
---
> <li class="toctree-l1 current"><a class="reference internal" href="Library.html">Library</a></li>
177c179,183
<       <li>N-Back Model (Beukers et al., 2022)</li>
---
>           <li><a href="Library.html">Library</a> &gt;</li>
>         
>           <li><a href="Models.html">Models</a> &gt;</li>
>         
>       <li>Nback Model</li>
200,268c206,213
<   <section id="n-back-model-beukers-et-al-2022">
< <h1>N-Back Model (Beukers et al., 2022)<a class="headerlink" href="#n-back-model-beukers-et-al-2022" title="Permalink to this headline">¶</a></h1>
< <p><a class="reference external" href="https://psyarxiv.com/jtw5p">“When Working Memory is Just Working, Not Memory”</a></p>
< <section id="overview">
< <h2>Overview<a class="headerlink" href="#overview" title="Permalink to this headline">¶</a></h2>
< <p>This implements a model of the <a class="reference external" href="https://en.wikipedia.org/wiki/N-back#Neurobiology_of_n-back_task">N-back task</a>
< described in <a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al. (2022)</a>.  The model uses a simple implementation of episodic
< memory (EM, as a form of content-retrieval memory) to store previous stimuli along with the temporal context in which
< they occured, and a feedforward neural network (FFN)to evaluate whether the current stimulus is a match to the n’th
< preceding stimulus (nback-level)retrieved from episodic memory.  The temporal context is provided by a randomly
< drifting high dimensional vector that maintains a constant norm (i.e., drifts on a sphere).  The FFN is
< trained, given an n-back level of <em>n</em>, to identify when the current stimulus matches one stored in EM
< with a temporal context vector that differs by an amount corresponding to <em>n</em> time steps of drift.  During n-back
< performance, the model encodes the current stimulus and temporal context, retrieves an item from EM that matches the
< current stimulus, weighted by the similarity of its temporal context vector (i.e., most recent), and then uses the
< FFN to evaluate whether it is an n-back match.  The model responds “match” if the FFN detects a match; otherwise, it
< either responds “non-match” or, with a fixed probability (hazard rate), it uses the current stimulus and temporal
< context to retrieve another sample from EM and repeat the evaluation.</p>
< <p>This model is an example of proposed interactions between working memory (e.g., in neocortex) and episodic memory
< e.g., in hippocampus and/or cerebellum) in the performance of tasks demanding of sequential processing and control,
< and along the lines of models emerging machine learning that augment the use of recurrent neural networks (e.g., long
< short-term memory mechanisms; LSTMs) for active memory and control with an external memory capable of rapid storage
< and content-based retrieval, such as the Neural Turing Machine (NTN;
< <a class="reference external" href="https://arxiv.org/abs/1410.5401">Graves et al., 2016</a>), Episodic Planning Networks (EPN;
< <a class="reference external" href="https://arxiv.org/abs/2006.03662">Ritter et al., 2020</a>), and Emergent Symbols through Binding Networks (ESBN;
< <a class="reference external" href="https://arxiv.org/abs/2012.14601">Webb et al., 2021</a>).</p>
< <p>The script respectively, to construct, train and run the model:</p>
< <ul class="simple">
< <li><p>construct_model(args):
< takes as arguments parameters used to construct the model; for convenience, defaults are defined toward the top
< of the script (see “Construction parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>train_network(args)
< takes as arguments the feedforward neural network Composition (FFN_COMPOSITION) and number of epochs to train.
< Note: learning_rate is set at construction (which can be specified using LEARNING_RATE under “Training parameters”).</p></li>
< </ul>
< <ul class="simple">
< <li><p>run_model()
< takes as arguments the drift rate in the temporal context vector to be applied on each trial,
< and the number of trials to execute, as well as reporting and animation specifications
< (see “Execution parameters”).</p></li>
< </ul>
< <p>The default parameters are ones that have been fit to empirical data concerning human performance
< (taken from <a class="reference external" href="https://psycnet.apa.org/record/2007-06096-010?doi=1">Kane et al., 2007</a>).</p>
< </section>
< <section id="the-model">
< <h2>The Model<a class="headerlink" href="#the-model" title="Permalink to this headline">¶</a></h2>
< <p>The models is composed of two <a class="reference internal" href="Composition.html"><span class="doc">Compositions</span></a>: an outer one that contains the full model (nback_model),
< and an <a class="reference internal" href="AutodiffComposition.html"><span class="doc">AutodiffComposition</span></a> (ffn), nested within nback_model (see red box in Figure), that implements the
< feedforward neural network (ffn).</p>
< <section id="nback-model">
< <h3>nback_model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h3>
< <p>This contains three input Mechanisms (</p>
< <p>Both of these are constructed in the construct_model function.
< The ffn Composition is trained use</p>
< <figure class="align-left" id="nback-fig">
< <img alt="N-Back Model Animation" src="_images/N-Back_Model_movie.gif" />
< </figure>
< </section>
< </section>
< <section id="training">
< <h2>Training<a class="headerlink" href="#training" title="Permalink to this headline">¶</a></h2>
< </section>
< <section id="execution">
< <h2>Execution<a class="headerlink" href="#execution" title="Permalink to this headline">¶</a></h2>
< <p>Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code>
< .. Script: <code class="xref download docutils literal notranslate"><span class="pre">N-back.py</span></code></p>
< </section>
---
>   <section id="nback-model">
> <h1>Nback Model<a class="headerlink" href="#nback-model" title="Permalink to this headline">¶</a></h1>
> <p><strong>“When Working Memory is Just Working, Not Memory”</strong> (<a class="reference external" href="https://psyarxiv.com/jtw5p">Beukers et al., 2022</a>)</p>
> <p>Script: <a class="reference download internal" download="" href="_downloads/3ca6d8ab14ce9c785a6b5c1b1d60a955/nback.py"><code class="xref download docutils literal notranslate"><span class="pre">nback.py</span></code></a></p>
> <p>Example of Jupyter Notebook: <a class="reference internal" href="BeukersNBackModel_NB.html"><span class="doc">Nback Model Notebook</span></a></p>
> <p>(Instructions for running the actual notebook can be found <a class="reference internal" href="index_logo_with_text.html#tutorial"><span class="std std-ref">here</span></a>, replacing <code class="docutils literal notranslate"><span class="pre">nback_nb</span></code> for <code class="docutils literal notranslate"><span class="pre">tutorial</span></code>)</p>
> <div class="toctree-wrapper compound">
> </div>
276a222,230
>     <div class="rst-footer-buttons" role="navigation" aria-label="footer navigation">
>       
>         <a href="ContributorsGuide.html" class="btn btn-neutral float-right" title="Contributors Guide" accesskey="n" rel="next">Next <img src="_static/images/chevron-right-orange.svg" class="next-page"></a>
>       
>       
>         <a href="BustamanteStroopXORLVOCModel.html" class="btn btn-neutral" title="LVOC Model of Stroop XOR Task (Bustamante et al. 2017)" accesskey="p" rel="prev"><img src="_static/images/chevron-right-orange.svg" class="previous-page"> Previous</a>
>       
>     </div>
>   
305,314c259
< <li><a class="reference internal" href="#">N-Back Model
...

See CI logs for the full diff.