Mental Representations and their Role in the Cognitive Sciences

Mental Representations and their Explanatory Role in the Cognitive Sciences

Since the start of the cognitive revolution, mental representations have played a central role in the cognitive sciences and the philosophy of mind.  It seems that every type of cognitive explanation – such as general philosophical theories, models in cognitive psychology, computational models, and neuroanatomical research – all use the notion of mental representation.  Yet there is no consensus over the nature of representation- What is the ontological status of mental representations? What is their explanatory role?  And how do the answers to such questions differ with regards to the different types of cognitive explanations?

A study group involving Ph.D. students from different disciplines of cognitive research (such as philosophy, cognitive psychology, linguistics, computational neuroscience, and neuroanatomy) will allow for a thorough and extensive discussion over the use of representations in the cognitive sciences.

Furthermore, the fact that mental representations are found in a wide variety of cognitive sciences makes it an ideal subject for creating a discussion between otherwise distinct types of research.  This is an opportunity for an interdisciplinary conversation in cognitive science, which might deepen our understanding of different cognitive explanations and the relations between them.

The group will meet once a month.  In each meeting, one member of the group will present her or his research.  We will then discuss the explanatory role of representations within this type of research, what this use of representations tells us, and how it relates to different types of research. 

ייצוגים מנטליים ותפקידם ההסברי במדעי הקוגניציה

מתחילתה של המהפכה הקוגניטיבית, ייצוגים מנטליים ממלאים תפקיד מרכזי עבור הסברים במדעים הקוגניטיביים ובפילוסופיה של ה-mind.  נדמה שכל רמה של הסבר קוגניטיבי - למשל תיאוריות פילוסופיות כלליות, מודלים בפסיכולוגיה קוגניטיבית, מודלים חישוביים, ומחקרים נוירואנטומיים - כולם משתמשים כך או אחרת ברעיון של ייצוג מנטלי.  למרות זאת, אין הסכמה ברורה לגבי אופיים של ייצוגים כאלו- מהו מעמדם האונטולוגי?  מה תפקידם ההסברי?  ובאיזה אופן, אם בכלל, התשובות לשאלות אלו משתנות בין דיסציפלינות שונות במדעי הקוגניציה?

קבוצת חברותא הכוללת דוקטורנטים שעוסקים בהסברים קוגניטיביים ממספר כיוונים שונים (כגון פילוסופיה, פסיכולוגיה קוגניטיבית, בלשנות, חישוביות עצבית, נוירואנטומיה), תאפשר דיון מקיף ומעמיק באופן השימוש של המדעים הקוגניטיביים בייצוגים מנטליים.

מעבר לכך, עצם העובדה שייצוגים מנטליים נוכחים במגוון רחב של מדעים קוגניטיביים הופכת נושא זה לאידיאלי על מנת לייצר דיון בין סוגי מחקר שונים.  זו הזדמנות לקיים מסגרת לשיח אינטרדיסציפלינרי במדעי הקוגניציה, שיכול להועיל להעמקת ההבנה של סוגי ההסברים השונים והקשרים ביניהם.

הקבוצה תיפגש פעם בחודש.  בכל מפגש, אחד המשתתפים יציג את המחקר שבו הוא עוסק.  לאחר מכן יתקיים דיון על התפקיד ההסברי של ייצוגים יחסית למחקר הזה, מה ניתן ללמוד מכך, ואיך זה מתקשר לשימושים אחרים של ייצוגים ולמחקרים אחרים עליהם דיברנו.


Group Meetings

Meeting 1

Date: 14.12.2016

Participants: Nora, Amnon, Alon, Yinon, Avi, Jonathan, Aviv, Noam, Galit, Ori

Presenter: Ori - On mental representations and this chevruta.


Participants introduced themselves, their research, and their interest in this group.

Ori presented an introduction of mental representations and his research interests regarding them.  We discussed the representationalist/eliminativist debate and mentioned the option of a pragmatic approach to representations.

The main takeaway from our discussion of representations was probably the simple understanding that the notion of representation in the cognitive sciences is both extremely important and very much unclear.

We set the general aim of the chevruta to better understand the role of mental representation in cognitive research.

We agreed to try and achieve that by looking at one theory at a time, which should help in grounding and focusing the discussion.  Thus the next meetings will include a presentation of a specific research/theory/model which will then be followed by a discussion over the role of representations with regards to that specific research.

The next meeting will be held on 4.1.2017.  Alon will present his research.

PowerPoint presentation attached. 

Meeting 2

Date: 11.1.2017

Participants:  Alon, Jonathan, Noam, Amnon, Aviv, Ori

Presenter: Alon- On Non-Conscious Representations



Alon presented some of his lab's research.  Abstract:

The traditional claim in cognitive sciences is that complex and serial cognitive functions can only be performed on consciously available representations.  This claim was tested many times in the past, always using short time frames, allowing stimuli (and therefore the mental representation following it) to remain subliminal.  Lately, however, using a new technology allowing us to render stimuli subliminal for longer periods, this barrier was broken.  In our meeting, Alon presented some of the original findings, supporting the claims of inability to process non-conscious mental representation, as well as some new paradigms from his lab, providing evidence of complex, serial and strategic processing of non-conscious mental representations.

PowerPoint presentation attached.

Alon's presentation led to various debates with regards to the role of mental representations:

The first question to arise was whether or not this type of research proves the existence of non-consious representations.  We seemed to agree that it proves the existence of non-concious high level cognitive capacities.  But with regards to the existence of non-concious representations, that seems dependant on whether or not we already assume that high level cognitive capacities (like the ability to calculate 15-7-2) can only be performed through a manipulation of representations.

This immediately led to a second debate- Is there truth to the claim that high level cognitive abilities can only be performed through the manipulation of representations?  This debate was not settled.  We tried to understand the justification behind such a claim- why must high-level functions demand the existence of representations?  Is it because that when I calculate 15-7-2 conciously it seems/feels like I'm manipulating representations (of 15,7, etc.)?  If so, then doesn't Alon's research, showing these tasks can be achieved non-conciously, prove that that is a weak justification?  Perhaps the justification in assuming the need for representations is simply derived from the complexity of these tasks.  If so, then does that mean that when calculaters or computers achieve such tasks, they too must manipulate representations?  And would that be the same type of representation manipulation that human beings are assumed to perform?

These last questions offer a quick stepping stone to the biggest questions with regards to representation.  Do machines or computers use representations like humans do?  What counts as a representation?  What counts as using a representation to achieve a task?  We discussed these questions shortly and we will surely return to them.  Another important question that was touched upon was an attempt to analyze the extent to which cognitive psychology is commited to representations.  Are cognitive psychologists realists, pragmatists, or maybe eliminativists?  And how do we even tell the difference?  We will return to this question as well (perhaps even right at the start of our next meeting).

Meeting 3

Date: 25.1.2016

Participants: Aviv, Alon, Galit, Noam, Ori, Avi, Jonathan

Presenter: Aviv


Aviv presented his work on mathematical objects and their cognitive representations.  Abstract:

What's considered to be objects in the realm of mathematics, are very different from what we usually take for objects, in real life. But to the extent that we do relate to them when doing mathematics, they too should be somehow represented in the mind. What kind of creatures are these representations? To what extent does the system managing them share substance with our ordinary objects representations management? Can we learn something about mathematical thinking from the regular cognitive functioning? Could we learn something about mathematics itself? Could anything be more fun than this?

The discussion following focused mostly on what being a mental representation is, the different kinds of representations, and what does it mean to say that there are no representations.  However, we didn't get very far in this discussion and we agreed to return to Aviv's work next time to try and gain more prespective on his research and allow for another discussion.  So- to be continued...

Meeting 4

Date: 15.2.2017

Participants:  Aviv, Jonathan, Yinon, Amnon, Ori

Presenter: Aviv


Abstract for Aviv's talk:  We thought we knew what numbers there are. We thought infinite ones don't make mathematical sense. We thought wrong, kinda. But the notion of numbers splits into two different ones ("Ordinals" and "Cardinals"). Which apparently, in the finite, are both integrated into one. Kinda. I will present the story, and offer a theoretical cognitive framework through which to understand it. The framework sets to account (in the most general terms) for how the cognitive system comes to learn new object-types - of all kinds (Apple, Cat, Physical object, Number, etc.). In such a framework, we could then begin to explore the learning of mathematical objects as on par with "ordinary" ones, and their place in our mental universe. The talk will generally be independent of the previous one, but just as fun. Kinda.

 Aviv's proposed cognitive framework is centered on the notion of a "procedure array" relevant to an object.  We discussed the extent to which Aviv's theory is committed to mental representations, if at all.  Aviv mentioned that he normally refers to these "procedure arrays" as (mental) representations, because it is a usefull way to present them, but it might be possible to avoid labeling "procedure arrays" as representations.  Since "procedure arrays" are, well, an array of (cognitive) procedures it could be possible to claim that they only as exist as a distinct array from the point of view of an outside theorist, therefore not serving as an internal representation for the system itself.  However, what if the system itself can access a "procedure array" as such?  Is that not reason enough to call this array an internal representation?  What does it mean for the system to access a "procedure array" of an object X as such?

Meeting 5

Date: 7.3.2017

Participants: Alon, Galit, Yinon, Ori

Presenter: Ori


The meeting had two parts.  The first was an attempt to classify some of the questions that have come up so far- with regards to mental representations.  We discussed two main views- on the first the commitment to representations is in effect a commitment to some cognitive phenomena and nothing more.  The second view is much stronger and is committed to the existence of actual (physical) internal representations as a part of the cognitive mechanism that underlies and enables the relevant phenomena.  We tried to better understand the extent of this commitment, and what it would take to make an argument for it.

In the second part of the meeting, Ori presented a possible claim that something can only function as a reprsentation within a mechanism if there is also a cognitive agent within that mechanism that uses it as such.  He claimed that if this is the case, the strong view on internal mental representations is no longer available (for it would require an "internal" cognitive agent.)  Arguments for and against these claims were discussed.

Meeting 6

Date: 28.3.2017

Participants:  Galit, Eva, Henry, Aviv, Ori

Presenter: Galit


This was a joint meeting, together with the langauge processing chevruta.  Galit presented her work on mental representations in langauge.  She began with some background on the semantics of representations and how they relate to different linguistic structures.  We discussed whether some sentence structures are more basic than others.

Specifically, a "straightforward" positive sentence seems to require less cognitive processing than a sentence that includes a negation or a passive action.  We discussed the claim that the added difficulty in these cases is a result of a more complicated syntactic structure rather than a semantic difference in the mental representation.  Galit presented her research into this claim while focusing on the case of negation, and how its syntactic structure requires added cognitive processing.