Investigating the effect of varying teaching conditions and chain positions on human cumulative culture
Investigating the effect of varying teaching conditions and chain positions on human cumulative culture.
Abstract:
The concept of cumulative culture in humans is one that has been researched by psychologists such as Christine Caldwell (2007) in order to discover how teaching can affect learning and how necessary it is. This previous research led to the focus of this experiment, which is to explore how teaching conditions and chain positions in groups can affect learning. Reported in this document are the results of a laboratory experiment testing these variables. Participants were asked to play a game of battleships on a computer, each under one of four teaching conditions. Results of how many ships each participant sunk were then recorded, along with which teaching condition the participant was in and the means and standard deviation were recorded. The results showed that learners in later chain positions scored higher than those in earlier ones however, surprisingly also showed that there is little conclusive evidence to support teaching conditions affecting participant’s scores. Finally, this report provides recommendations for future research on how learning differs between genders.
Introduction:
Cumulative culture is the process of social transmission of information from one individual to another. This has been a topic of interest for psychologists in order to establish whether teaching and cumulative culture are linked. This is because such links could explain how humans respond to teaching and successfully pass down knowledge. Research papers done by Christine Caldwell, Mark Atkinson and Elizabeth Renner such as, ‘Experimental Approaches to Studying Cumulative Cultural Evolution.’ (2016) inspired the basis for this experiment in order to build upon what they have originally found. Within their reported experiment they found that participants in later chain positions performed better than those in earlier positions and that when given the opportunity to observe earlier groups, performed better still. However, there is also evidence to suggest that teaching may not actually be required for cumulative culture trends to occur (Human Teaching and Cumulative Cultural Evolution, p3). Tomasello, Kruger and Ratner (1993) described human culture as working in a similar fashion to that of a ratchet, whereby progress increases in functionality over generations and rarely goes backwards, suggesting that chain position scores should increase as the chain positions do. This previous research inspired the experiment that this report focuses on. It focuses on two main hypotheses, firstly, that the last participants in chains will score higher than the first if asked to participate in a game testing cumulative culture and secondly, final learners in chains who can communicate with teachers will perform better than those who cannot. The previous research led to the belief that these hypotheses may be correct. On the other hand, the null hypotheses of the experiment were that chain position (hypothesis one) and teaching conditions (hypothesis two) would have no effect on success within the experiment.
Methods:
The experiment was composed of 243 participants- 199 of which females and 44 male. These participants were all students at the University of Stirling studying Psychology and completing the experiment as a coursework task in October 2018. They were divided into age categories which resulted in 224 participants aged 16-25, 14 participants aged 26-35 and 5 aged 36-45. Ethical approval was sought and gained from the University of Stirling GUEP Ethics Committee and consent gained from all participating individuals. The participants were broken up into 54 different microsocieties- small groups of learners that mimicked generational succession by removing and then replacing the participants with new groups after the completion of the task. This process created the chain groups. The groups were asked to come to a computer room at an allocated time previously arranged. Upon arrival, participants were led to a desktop computer to sign a consent agreement of participation after reading the ethics and information guidelines of the experiment. They were also given a unique number in order to preserve anonymity that they could use if they decided to withdraw from the study, therefore withdrawing their consent to use their data. The participants were anonymous and none continued to complete the experiment without providing consent. After doing so, they were led to a different desktop computer where a computerised game of battleships was waiting for them. This was provided by using a grid-search task ran in PsychoPy (Pierce, 2009).
Design:
The hypotheses were tested using a laboratory based experiment under controlled conditions in the form of a transmission chain study. The independent variables were the chain position of the participant and teaching conditions of the participant. The dependant variable was the number of ships participants were able to sink in a computerised game of battleships. This was similar to the setting of experiments carried out by Caldwell and Millen (2008) who also demonstrated cumulative culture evolution in laboratory conditions using microsocieties completing a task. It was predicted that those in a later chain position would perform better at the task than those in the earlier chains and that those who received information from the teacher would score higher than those without help. Therefore, the variable used to measure the success of the experiment was the number of ships successfully sank by the participants.
Procedure:
As previously mentioned, participants started the experiment by arriving in small groups to a computer room with seating available for the next group to wait before beginning the experiment themselves. Three members of staff/assistants were present during the experiment. The participants were each given a number and instructed to take a seat in the waiting area before beginning. One by one, they were called to the desktop computer, so they could read through the consent forms and study brief and decide whether or not to participate. They were then led to the other desktop computers provided, already set up with the grid-search task programme and given time to practice the task before beginning properly. After being asked and confirming that the participants were comfortable to begin, an assistant would start the real experiment involving three grids containing nine hidden ships within per grid. They were allowed 36 attempts to hit ships per grid including four search attempts and nine misses allowed per attempt. If the participant came second, third or fourth place in the chains, earlier members of the chain send information to every ninth ‘miss’ in order to assist them in sinking ships.
There were four teaching conditions within the experiment;
1. Intentional feedback- where the learner selects grid squares revealed during their search to send to the teacher.
2. Grid squares revealed in the learners search for ships are randomly selected by the computer and then send to the teacher
3. Full feedback- all of the grid selections made by the learner are sent to the teacher meaning the teacher can access all performance history from the learner.
4. No feedback- information is transferred from teacher to learner leaving the teacher with no information about the learner’s search.
The aim was to study whether participants in one teaching condition scored significantly higher or lower than those in others.
The data was then collected and programmed into an excel document with the total score per participant inserted into it. The data was then categorised into groups showing which chain each participant was in, which condition of teaching they had and their score by chain position. Then the mean scores and standard deviation were calculated in order to discover trends in the findings as demonstrated in Figures one to four. Descriptive statistics such as means, standard deviation and T-tests were then carried out in order to determine whether the mean significantly differed from the expectation of results and whether or not the hypotheses were supported.
Results:
The mean number of ships sunk (including Standard Deviation) for all four chain positions within the experiment are displayed in Figure 1. This shows that, as predicted by hypothesis one, there is a positive correlation between number of ships successfully sunk and chain position.
Figure 2.
Table 1:
Mean Scores and Standard Deviation of Chain Position Data
Score By Chain Position
1 2 3 4
Mean Score 9.222 12.227 14.148 14.351
SD 2.625 2.594 3.080 3.239
Fig. 2 shows a table of mean scores and standard deviation per chain position. Reflected in fig. 1, it shows a general increase from position one to four despite one slight anomaly (the standard deviation for chain position one). This suggests that with each chain position, the number of ships sunk would increase, showing evidence of cumulative culture and perhaps imitation within the experiment.
Furthermore Figure 3 shows a general increase again, when chain position increases. However, the results do not seem to suggest a definite trend towards better scores with more teaching.
T-Test 1:
T-test one, compared chain groups one and four putting the null hypothesis to the test by seeing whether the difference differed from zero. The mean score for first chain position participants was 9.22 with a standard deviation of 2.63 while those in chain four achieved much higher scores with a mean of 14.35 and a standard deviation of 3.24. The t(fd)=53.00, t-test value =-8.7037699794525100 and P-value= 0.0000000000085521. This shows limited support for the null hypothesis therefore confirming the first hypothesis to be correct, suggesting that participants in later chains score higher than those in beginning chains.
Table 2:
Table Showing Standard Deviation of Teaching Conditions in Chain Positions
CP 1 CP 2 CP 3 CP 4
Full Information 2.106157 1.898042 3.010665 3.391165
Inadvertent Information 2.278664 3.692136 3.860194 2.939874
Intentional Information 3.383866 2.353394 3.130846 4.127102
No Information 2.660249 1.921538 2.415229 2.192645
Note: ‘CP’= Chain Position
T-test two analyses the data in order to show whether hypothesis two was correct by comparing different teaching conditions. Fig. 4 shows the standard deviation of the scores per teaching group. These figures appear to support the null hypothesis -that there is no link between teaching condition and higher scoring. The T(df)= 26.00, the t value= 0.15 and the P value= 0.87. This result suggests that it cannot be concluded that a significant difference between teaching conditions exists, therefore confirming the null hypothesis for hypothesis two.
Discussion:
Hypothesis one (that the last participants in chains will score higher than the first if asked to participate in a game testing cumulative culture) was supported by the data gathered in the experiment showing a positive correlation. Hypothesis two (that final learners in chains who can communicate with teachers will perform better than those who cannot) was found to have little evidence to support it, disproving the hypothesis. This is supported by psychological literature by Caldwell et al (2017) whereby it is suggested that ‘”Teaching” may not necessarily function to facilitate learning’ (Human Teaching and Cumulative Cultural Evolution, p.3). This reinforces the theory suggested in psychological literature by Lewis and Laland (2012) that imitation and copying may be more important to successful learning than teaching.
It is important to acknowledge potential confounding variables such as those which may have reduced the reliability or validity of the results for example, participants rushing to finish their games and the difference in ratio of female to male participants. It could be hypothesised that differences in gender may have an effect on how participants react to different teaching conditions. Furthermore other factors such as participant’s concentration levels throughout the experiment may have affected the results, for example if the participants were unfocused due to hunger or tiredness, they may score lower scores than those who are not.
In conclusion it is clear from the results of the experiment that chain position does affect the success of participants when faced with tasks but that teaching condition does not necessarily have a significant impact. Leading on from this research, further possible research opportunities may lie in considering whether or not gender effects the way participants react to teaching conditions. This would allow more insight into gender differences in learning behaviours.
