Visual attention

I started reading an article “Computational modeling of visual attention“, by Itti & Koch, 2001. First conclusion: ‘perceptual saliency of stimuli critically depends on the surrounding context’. It deals with the idea of a (single) lady in a red dress.

Further reading on centre-surround mechanisms, brought me to this news article: “Visual Attention: How The Brain Makes The Most Of The Visible World” (ScienceDaily 2011), based on two articles of Sundberg, Mitchell and Reynolds, plus Reynolds and Heeger.

(Credit: Salk Institute for Biological Studies)

“When we instead direct our attention to a stimulus in the surround — the violinist let’s say — the neuron’s response to the now irrelevant bassist is suppressed.”

I response to the neuron-metafor used in my previous post, let’s do a thought experiment and quote the the news article and replace specific words:

Persons in the visual cortex group view the world through their “receptive fields,” the small portion of the visual field individual neurons persons actually “see” or respond to. Whenever a stimulus falls within the , the cell group member produces a volley of electrical spikes thoughts and/or responses, known as “action potentials” that convey information about the stimulus in the receptive field.
But the strength and fidelity of these signals also depends on other factors. Scientists generally agree that neurons persons typically respond more strongly when attention is directed to the stimulus in their receptive fields. In addition, the response of individual neurons individuals can be strongly influenced by what’s happening within the immediate surroundings of the receptive field, a phenomenon known as contextual modulation.
“The surround has the ability to suppress the neuron’s persons response,” explains first author Kristy Sundberg, Ph.D., a former graduate student in Reynolds’ lab and now a postdoctoral researcher at Yale University. “It keeps us from responding all the time if there’s something that’s big and uniform and not particularly interesting or useful. This raised the possibility that the receptive field surround might provide a way to suppress the responses of task-irrelevant distracters.”

The text above seems to describe one previously found metafor.

Reading further on computational modeling, Itti and Koch state:

So, whereas certain features in the visual world automatically attract attention and are experienced as ‘visually salient’, directing attention to other locations or objects requires voluntary ‘effort’. Both mechanisms can operate in parallel. {Luckily we are not blind to the world outside our attention span}

Attention {1} allows us to break down the problem of understanding a visual scene into a rapid series of computationally less demanding, localized visual analysis problems. In addition to these orientating and scene analysis functions, attention {2} is also characterized by a feedback modulation of neural activity for the visual attributes and at the location of desired or selected targets. This feedback is believed to be essential for binding the different visual attributes of an object, such as colour and form, into a unitary percept. By this account, attention {3} not only serves to select a location of interest but also {4} enhances the cortical representation of objects at that location. As such, focal visual attention has been compared to a ‘stagelight’, successively illuminating different players as they take centre stage. Finally, attention {5} is involved in triggering behaviour, and is consequently intimately related to recognition, planning and motor control.

Visual attention is a principle also used by chess players, recognizing game states. However, when in a meeting, I asume groupmembers are not consiously aware of some sort of mindmap at hand. Two questions arise: (1) do we use a similar principle -like visual attention- when dealing with a narrative or a more abstract dialogue? Or don’t we? (2) What is the relation of principles known as ‘association’ and ‘priming’ to the constructs of visual attention?
Itti & Koch focus on focal bottom-up attention, and not on top down, volitional component of attention. If I understand this correct, this latter aspect seem scloser related to the second question regarding ‘association’ and ‘priming’.
All together, Itti & Koch describe (a) the dependance of saliency on context, (b) the idea model of a salience map, (c) the principle of Inhibition of return, (d) an ‘control/coordinate’ interplay of attention and eye-movement and (e) how sceneunderstanding and object recognition constrain selection of attention.



Dendrites and axons

After reading a book for children and parents about the learning states of a brain (dutch: Breinlink), I realized I forgot to share an interactive sketch which resembles our beloved axons and dendrites.

We start with two persons, who have a dialogue. In this dialogue, the experienced reach and the effective reach are not the same. In the middle we find the effective reach. The experienced reach is highlighted from the red persons persective (first image) and from the yellow persons perspective (second image):



Whether or not this difference between reach stays the same when adding members to a dialogue is questionable:



Suppose we have some interaction in a group of five. Each dotted line represents an interaction between 2 groupmembers. The rosettes in the outer ring represent the interactions per person. The pentagon in the middle is an overlay of all interactions between the 5 persons. Only the reach of the red person is highlighted:


Let’s also suppose these persons all work for different organizations. They all have an organizational backbone. Let’s draw the interactions with these backbones as well:


The picture starts to resemble a small system of brain cells, each with its dendrites and axons, doesn’t it? Here’s a really short video:


Wanna play yourselves? Here’s the applet. Make sure your browser supports Java.


Complexity, Energy & time

A news item of the Santa Fe Institute led me to this talk of Geoffry West:

At 15min50, there’s a slide telling us that:

… organisms have evolved by natural selection,
so as to:
a) minimize energy dissipated in the networks
b) maximize the scaling area of interface with their recource environment

From the domain of this research we could hypothesize:

… organizations have evolved by natural selection,
so as to:
a) minimize energy dissipated in the networks
b) maximize the scaling area of interface with their recource environment

What does this tell us about the interaction between cells, organisms, people and organizations? What happens at the interface of a mobile organism in between organizations? This is an area where people interact, balancing between the a)’s and b)’s of different organizations.

At about 22min West explains:

‘If I take time …and if I correct {some stuff} …everybody lives, evolves, dies at the same rate …if you do the appropriate corrections’.

This helps us realize, and not define, the timing difficulties in complex decision making, lifted over the walls of standing organizations.

In a similar SFI video D’Souza (at 28min) explains that – when using a little bit of communication – regarding to non zero sum games:

Coordination brings USER optimal in line with SYSTEM optimal


Different types of decision trees

After reading an article in my newspaper about decision making and the role of regulators (dutch, source NRC) plus this older article (dutch, 1999, source nrc), I found two different types of decision trees.

First, there is one like Prof. Dr. Sweder van Wijnbergen developed for the dutch ministry of finance. This tree is more a decision making flow chart (source aef, page 11), a tree that leads to action based on a framework of experience:


Second, there is the type that helps compare possible outcomes (source wiki), that leads to action based on ball park figures and best estimates:


Both decision trees however offer the support of shared understanding. This overview is not only helpfull to work towards the decision itself, but mainly helps exploring and explaining the process of decision making. They support a dialogue better and faster that any text paragraph would do and offer support in conflict resolution.

There’s clearly a functional overlap between decision trees and argument maps (see periodic table, information visualization). All types have both a sense making function and a memory / retention function.
This slidedeck
explains more about how elements of a tree/map support organizing information: segmentation (a series of propositions), coding & hierarchy.


Political interruptions

I am intrigued by the clumsy graphical translation in my paper today (NRC weekend, september 24th and 25th 2011, volume 303 Algemeen Handelsblad), about the dutch annual policy reflections in The Hague:


1) upper left: what a mess. Why not use a rosette system like this one?
2) upper right: why let these colors compete for attention with the small bandwidth of widths of the arrows?
3) lower left: why not write: ‘People who have opposite standpoints interrupt each other more often’?

Luckily there is a source: Political mashup: research about wordclouds and xml by the Amsterdam University. The paper and Political mashup collaborate and have a shared platform with conceptual graphics.
Btw, this original image shows better proportions of arrow widths.

Diging a little bit deeper we find an amusing and interesting student contribution to an impressive Open Data Challenge (2011):


From there, through the red tab, we find the orignal funny and mature looking source of the clumsy translation by my newspaper:


… I was a bit puzzling finding out how the color coding works, though. Very well done, Jurrian Tromp, Reinier van der Plank en Thomas Moeskops! They appear to have removed quite some distracting elements from one of their inspirations: I hope Political Mashups and NRC keep learning and perhaps sponsoring students or providing internships! Perhaps this will one day lead to visual support of understanding political profiles and voting.



An inspiring animation (Balance, 1989, by Wolfgang & Cristoph Lauenstein) about group balance:

… found it as part of a ‘Tegenlicht’ episode (VPRO, September 19th 2011): “Shortly after the start of the 2008 financial crisis sociologist Manuel Castells gathers a small group of international intellectuals around on a deeper refelection regarding the crisis”.