I am not aware about any recent publications about this precise question, but I can imagine recollection being faster because the association of the cue (i.e. on screen) and the memory it triggers is stronger and therefore easier accessible than the one of familiarity (where obviously ppl are not that sure of the source memory of an object/picture, etc.). Would be interesting to find out any recent evidence on this. 

Hi Dorian,

One possibility that has been discussed is also that recollection/remember responses tend to be associated with higher confidence than familiarity/know responses. In a single-process account, higher confidence would be associated with faster RTs.

http://www.psych.umass.edu/uploads/sites/48/Files/Rotello%20Zeng%202008.pdf

This is similar to what Clarissa suggests with the caveat that if you control for confidence, you may be able to remove the RT difference.

Thank you for your answers. I am aware of the degree-of-confidence hypothesis and the long lasting debate on the existence of recollection and familiarity. In fact, my question was stimulated after seeing some recent papers still published on this topic by Squire and/or Wixted. When I did my experiments some years ago, I found that activation of the medial temporal lobe was very much dependent on reaction times. Yet, people seem to give for granted that recollection and familiarity take different times and focus on assigning their role to substructures of the temporal lobe. In theory familiarity should be faster (according to Yonelinas and Jacoby 1994, who used speeded recognition tasks), but in practice recollection is found to be almost always faster (Rotello, as cited above, etc). I am still not aware how this is interpreted in the dual-process framework, or perhaps I am not sufficiently updated, that's why I asked.

Thanks for any input.

I remember writing about this in my PhD thesis, but I am a bit fuzzy about the details at the moment. Steve Dewhurst has done some work on this, and he may have a dual-process take on the issue:

http://eprints.lancs.ac.uk/859/2/Dewhurst,_Holmes,_Brandt,_&_Dean.pdf?origin=publication_detail

Hi all, I recently had a couple of papers out on recollection and familiarity RTs but in artificial grammar learning rather than recognition memory.

http://www.lifesci.sussex.ac.uk/home/Zoltan_Dienes/Mealor%20&%20Dienes%202012%20no%20loss%20gambling.pdf

http://www.lifesci.sussex.ac.uk/home/Zoltan_Dienes/Mealor%20Dienes%202013%20Speed%20of%20metacognition.pdf

In the latter paper we asked participants to classify sequences on the basis of guessing, familiarity-like or recollection-like responses while controlling for confidence. Recollection-based responses were still faster than familiarity-based ones when accounting for this. In Expt 2 we introduced response deadlines, which reduced the number of recollection-like responses in favour of guessing but affected the quality of familiarity-like responses (but performance was still above chance). We argued that when recollection is not forthcoming, people tend to probe their familiarity to get the most from it but it can still be elicited rapidly if forced. 

Thank you for participating. In sum, there is a reduced number of recollection responses when there is a response deadline, but recollection responses are faster when there is no deadline. Would this mean that people use different strategies depending on the allowed time? This would also fit with the Geraci studies who showed that small modifications of the remember/know instructions produce different results. And thus, these processes (if they exist) are very much dependent upon the conditions of the task: the deadline, the type of material, the levels-of-processing. In other words, we set the mind to a determinate encoding or retrieval mode. In this regard, if I remember correctly there was a paper from Jacoby 2005 showing that incidental encoding uses the level of processing being used at the moment. If you are trying to retrieve material at a deep level of processing (i.e., semantic) you also incidentally encode lure items in deep mode, and these items will be remembered better than those that are incidentally encoded in shallow mode.

My understanding is that dual-process memory models need to keep all these factors in consideration. And reaction times still remain partly unexplained, with some saying they fit the dual process model, while others saying they don't.

Yes I would think people approach these tasks differently depending on the task demands. In our results, on the trial-level, people actually shifted from recollection-based down to guessing on the basis of the deadline, rather than down to familiarity-based. (The exact terminology wasn't recollection and familiarity BTW). So it didn't appear that these were responses were ascribed to familiarity which were destined to become recollection if given enough time. In this sense, the results might actually fit well with the dual process signal detection models of the 90's where recollection is like an all-or-nothing threshold procedure and familiarity conforms to SDT. However, as these papers are on AGL rather than RK, they might not be so applicable to e.g. levels of processing  manipulations. Having said that, there is a paper by Turner and Fischler 1993 where people were or were not instructed to search for rules during AGL training (thought to increase explicit knowledge); when deadlines were introduced, this group's performance suffered the most, which is in line with slower responses for recollective type responses (but subjective reports were not used in this study).

Some papers that may help:

Contrasting Networks for Recognition Memory and Recency Memory Revealed by Immediate-Early Gene Imaging in the Rat . Olarte-Sánchez, Cristian M.; Kinnavane, Lisa; Amin, Eman; Aggleton, John P.. Behavioral Neuroscience, suppl. 4 128.4 (June 16, 2014): 504-522.

Dissociations in the effect of delay on object recognition: Evidence for an associative model of recognition memory. Tam, Shu K. E.; Robinson, Jasper; Jennings, Dómhnall J.; Bonardi, Charlotte. Journal of Experimental Psychology: Animal Learning and Cognition 40.1 (Jan 2014): 106-115.

The attentional boost effect with verbal materials. Mulligan, Neil W.; Spataro, Pietro; Picklesimer, Milton. Journal of Experimental Psychology: Learning, Memory, and Cognition 40.4 (Jul 2014): 1049-1063.

Relating the content and confidence of recognition judgments. Selmeczy, Diana; Dobbins, Ian G.. Journal of Experimental Psychology: Learning, Memory, and Cognition 40.1 (Jan 2014): 66-85.

Mammilliothalamic Tract Lesions Disrupt Tests of Visuo-Spatial Memory . Nelson, Andrew J. D.; Vann, Seralynne D.. Behavioral Neuroscience, suppl. 4 128.4 (June 23, 2014): 494-503.

The role of belief in occurrence within autobiographical memory. Scoboria, Alan; Jackson, Dennis L.; Talarico, Jennifer; Hanczakowski, Maciej; Wysman, Lauren; et al. Journal of Experimental Psychology: General 143.3 (Jun 2014): 1242-1258.

Oral approach–avoidance: Affective consequences of muscular articulation dynamics. Topolinski, Sascha; Maschmann, Ira Theresa; Pecher, Diane; Winkielman, Piotr. Journal of Personality and Social Psychology 106.6 (Jun 2014): 885-896.

Assume recall is a correlation process, equivalent to a matched filter to use signal processing terminology.  A given stimulus (image, sound, etc.) is applied to the set of matched filters.  Filters that are more closely tuned (through prior experience) to the stimulus will result in a relatively large correlation response.  Those with a close match (familiar) will have a lower response.  It stands to reason that the neurons wired to receive the responses from these correlators will be triggered (i.e. generate an action potential) sooner because the strongly correlated stimulus causes the detection threshold to crossed sooner.

In regard to your original question (and the task that was used), it has been suggested that faster recollection responses using the Remember/Know procedure may be an artifact of task instructions (Yonelinas, 2002). Often, participants must first assess recollection before they can determine whether a "know" response is required, because task instructions often require "know" responses to be made only when there are no recollective details available.

A paper that may be of some use is by Diana et al. (2006), which discusses the time course of recognition processes in regard to single and dual-process accounts: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2387212/

I noticed a recent paper in Neuropsychologia which provides further insight into the topic:

The posterior parietal cortex: Comparing remember/know and source memory tests of recollection and familiarity

http://www.sciencedirect.com/science/article/pii/S0028393214001924

The interesting finding is that different procedures (remember/know vs. source memory) have a different impact on the activation of the dorsal and ventral parietal cortices. It is interesting to note that reaction times were also different between responses of each probing method, though the authors have regressed out RTs through parametric modulations. Yet, the the dorsal to ventral shift from familiarity to recollection reminds me of similar effects of reaction times when I studied RTs a while ago. The crucial point of the study is, however, that different methods of recollection testing produce different activations in the parietal lobe, which gives a further push toward considering recollection as dependent on instructions and strategy, and perhaps also related to the time required to process with each strategy.

I believe Diana, Reder, et al. (2006) argued that recollection is always attempted first.

A simpler explanation is that the dual process model is wrong and that remember/know responses just index different degrees of memory strength. Many papers have found evidence that is congruent with this hypothesis (Donaldson, 1996; Wixted & Stretch, 2004; Dunn, 2004, 2008). That is, there is a single continuum of memory strength and remember responses merely represent responses from a higher decision criterion than know responses.