Merge pull request #94 from jeremymanning/master

more JRM edits

code/demographics.ipynb

@@ -1424,9 +1424,9 @@
  ],
  "metadata": {
   "kernelspec": {
-   "display_name": "attention-memory",
+   "display_name": "attention-memory.venv",
    "language": "python",
-   "name": "attention-memory"
+   "name": "python3"
   },
   "language_info": {
    "codemirror_mode": {
@@ -1438,7 +1438,7 @@
    "name": "python",
    "nbconvert_exporter": "python",
    "pygments_lexer": "ipython3",
-   "version": "3.10.10"
+   "version": "3.10.9"
   }
  },
  "nbformat": 4,

paper/main.tex

@@ -54,14 +54,18 @@
 important is happening there) or we may attend to particular features
 irrespective of their locations (e.g., when we search for a friend's face in a
 crowd versus a desired item in a grocery store). We ran a covert attention
-experiment with two conditions that differed in how long they asked
-participants to maintain the focus of the categories and locations they were
-attending. Later, the participants performed a recognition memory task for
-attended, unattended, and novel stimuli. Participants were able to shift the
-location of their covert attentional focus more rapidly than they were able to
-shift their focus of covert attention to stimulus categories, and the effects
-of location-based attention on memory were longer-lasting than the effects of
-category-based attention.
+experiment with two conditions that asked participants to either sustain or
+vary the focus of the categories and locations they were attending. Later, the
+participants performed a recognition memory task for attended, unattended, and
+novel stimuli. Participants in both conditions recognized attended stimuli more
+readily than unattended or novel stimuli. Participants in the ``sustained
+attention'' condition also exhibited a recognition advantage for
+attended-category images from the unattended location, whereas participants in
+the ``variable attention'' condition exhibited a recognition advantage for
+attended-location images from the unattended category. Our findings suggest
+that covert attention enhances memory encoding of attended stimuli, and that
+different aspects of partial attention affect memory encoding over different
+timescales.
 
 \noindent\textbf{Keywords: covert attention, volitional attention,
 location-based attention, category-based attention, recognition memory}
@@ -485,28 +489,29 @@ \section*{Results}
 Splitting participants' responses to face versus place images also revealed
 that participants often rated attended (and partially attended) place images as
 more familiar than attention-matched face images (compare dark versus light
-bars in Fig.~\ratingsByCategory). We hypothesized that this might be
-explainable by some property of the relevant cognitive processes or by
-properties of the stimuli themselves. To help elucidate this distinction, we
-examined individual exemplars of the face and place images used in our paradigm
-(Fig.~\ref{fig:stimuli}A). By design, the face images had consistent head
-sizes, viewing angles, expressions, and so on. In contrast, the place images
-varied more substantially across images. For example, some place images
-depicted human-made structures; others depicted natural scenes; some depicted
-indoor views; others depicted outdoor views; etc. This can also be seen by
-averaging the pixel intensity values across images, separately for the face and
-place stimuli (Fig.~\ref{fig:stimuli}B). Whereas the average face image retains
-many of the landmarks characteristic of most faces (e.g., clearly defined hair,
-eyes, nose, mouth, head shape, etc.), the average place image does not show
-place-specific features as clearly, aside from a general tendency for the tops
-of place images to be lighter than the bottoms of place images. We also
-computed the pairwise similarities across images from each stimulus category
-(Fig.~\ref{fig:stimuli}C) and found that face images tended to be much more
-similar to each other than place images (Fig.~\ref{fig:stimuli}D; $t(115258) =
-254.764, p < 0.001$). This analysis indicated to us that our experimental
-paradigm was not well-suited to identifying cognitively meaningful stimulus
-category differences, since participants' category-specific judgements may be
-confounded with within-category image similarity differences.
+bars in Fig.~\ratingsByCategory; sustained attention: $t(29) = 1.746, p =
+0.091$; variable attention: $t(22) = 2.660, p = 0.014$). We hypothesized that
+this might be explainable by either some property of the relevant cognitive
+processes or by properties of the stimuli themselves. To help elucidate this
+distinction, we examined individual exemplars of the face and place images used
+in our paradigm (Fig.~\ref{fig:stimuli}A). By design, the face images had
+consistent head sizes, viewing angles, expressions, and so on. In contrast, the
+place images varied more substantially across images. For example, some place
+images depicted human-made structures; others depicted natural scenes; some
+depicted indoor views; others depicted outdoor views; etc. This can also be
+seen by averaging the pixel intensity values across images, separately for the
+face and place stimuli (Fig.~\ref{fig:stimuli}B). Whereas the average face
+image retains many of the landmarks characteristic of most faces (e.g., clearly
+defined hair, eyes, nose, mouth, head shape, etc.), the average place image
+does not show place-specific features as clearly, aside from a general tendency
+for the tops of place images to be lighter than the bottoms of place images. We
+also computed the pairwise similarities across images from each stimulus
+category (Fig.~\ref{fig:stimuli}C) and found that face images tended to be much
+more similar to each other than place images (Fig.~\ref{fig:stimuli}D;
+$t(115258) = 254.764, p < 0.001$). This analysis indicated to us that our
+experimental paradigm was not well-suited to identifying cognitively meaningful
+stimulus category differences, since participants' category-specific judgements
+may be confounded with within-category image similarity differences.
 
 
 \begin{figure*}[tp]
@@ -667,9 +672,14 @@ \section*{Results}
 sustained attention condition do show some response biases. For example,
 participants tended to rate novel images as more familiar if they came from the
 just-cued category versus the unattended category (Fig.~\responseBias A; $t(29)
-= 4.371, p < 0.001$). Responses biases are more difficult to evaluate in the
-variable attention condition. For example, should cue recency be defined as the
-number studied composite image pairs that came between the image whose
+= 4.371, p < 0.001$). Nonetheless, participants' familiarity ratings in the
+sustained attention condition cannot be explained solely by response biases.
+For example, participants rate attended-category targets (i.e.,
+attended-category images they encountered at the unattended location) as more
+familiar than attended-category lures (i.e., attended-category novel images);
+$t(29) = 2.645, p = 0.013$. Responses biases are more difficult to evaluate in
+the variable attention condition. For example, should cue recency be defined as
+the number studied composite image pairs that came between the image whose
 familiarity the participant is judging and the most recent same-category cue
 (i.e., temporal distance to the nearest same-category attention cue)? Or might
 response biases instead arise when a given category is cued more often near the
@@ -751,6 +761,8 @@ \section*{Discussion}
 why unattended-category images at the attended location are rated as
 \textit{less} familiar in the sustained attention condition.
 
+
+
 The notion that location-based attention operates at a faster timescale than
 category-based attention is supported by prior work on the deployment of visual
 attention~\citep{SotoBlan04, StopEtal07}. Our findings that location-based
@@ -760,14 +772,15 @@ \section*{Discussion}
 category-based attention~\citep[e.g.,][]{MoheEtal14}. Our finding that people
 better remember attended stimuli also follows prior work on interactions
 between attention and memory~\citep{PallWagn02, ChunTurk07, AlyTurk16,
-AlyTurk17, WittEtal18, MorrEtal14, BaleEtal21}. Whereas much of this prior work
-focused on elucidating the neural basis of these interactions, our work extends
-these prior studies by elucidating the specific and separable behavioral
-impacts of location-based attention (enhancement with a fast onset) and
-category-based attention (suppression with a slow onset) and on subsequent
-recognition memory. Both of these effects persisted throughout the 2~min memory
-phases of both conditions. However, future work is needed to elucidate the
-longevity of these effects beyond 2 minutes.
+AlyTurk17, WittEtal18, MorrEtal14, BaleEtal21, UncaEtal11, TurkEtal13,
+LaRoEtal15}. Whereas much of this prior work focused on elucidating the neural
+basis of these interactions, our work extends these prior studies by
+elucidating the specific and separable behavioral impacts of location-based
+attention (enhancement with a fast onset) and category-based attention
+(suppression with a slow onset) and on subsequent recognition memory. Both of
+these effects persisted throughout the 2~min memory phases of both conditions.
+However, future work is needed to elucidate the longevity of these effects
+beyond 2 minutes.
 
 Another important area for future study concerns how the flow of information
 between different brain structures is modulated according to the focus of
@@ -783,20 +796,21 @@ \section*{Discussion}
 thalamus~\citep{Schn11}, whereas location-based attention may be supported by
 changes in connectivity with primary visual cortex~\citep{NoudEtal10}. That
 category-based and location-based attention are mediated by different brain
-structures may explain why these different aspects of attention operate on
-different timescales and affect memory differently. A strong test of this
-hypothesis would entail directly measuring neural activity patterns as people
-modulate their focus of attention (e.g., using functional magnetic resonance
-imaging or electroencephalography), and then using neural decoding
-approaches~\citep[e.g.,][]{HaxbEtal01, NormEtal06b, MannEtal18, OwenEtal21} to
-follow how neural representations of attended (or unattended) stimuli are
-transfered from primary sensory regions, to higher order sensory regions, to
-memory areas. If the effects of attention on memory are mediated by changes in
-network dynamics, the transmission rates of the representations of attended
-stimuli from primary sensory regions to memory areas should be facilitated
-relative to the transmission rates of unattended stimuli. Further, variability
-in these neural changes (e.g., as a participant focuses their attention with
-more or less success) should track with behavioral measures of memorability.
+structures~\citep[e.g.,][ and others]{GiesEtal03} may explain why these
+different aspects of attention operate on different timescales and affect
+memory differently. A strong test of this hypothesis would entail directly
+measuring neural activity patterns as people modulate their focus of attention
+(e.g., using functional magnetic resonance imaging or electroencephalography),
+and then using neural decoding approaches~\citep[e.g.,][]{HaxbEtal01,
+NormEtal06b, MannEtal18, OwenEtal21} to follow how neural representations of
+attended (or unattended) stimuli are transfered from primary sensory regions,
+to higher order sensory regions, to memory areas. If the effects of attention
+on memory are mediated by changes in network dynamics, the transmission rates
+of the representations of attended stimuli from primary sensory regions to
+memory areas should be facilitated relative to the transmission rates of
+unattended stimuli. Further, variability in these neural changes (e.g., as a
+participant focuses their attention with more or less success) should track
+with behavioral measures of memorability.
 
 Which aspects of our ongoing experiences we choose to attend affects how we
 process and remember those experiences later. Different forms of